ABSTRACT 
 
 
 
 
Title of Dissertation: DESIGN, FABRICATION, AND 
PERFORMANCE CHARACTERIZATION OF 
MULTIFUNCTIONAL STRUCTURES TO 
HARVEST SOLAR ENERGY FOR FLAPPING 
WING AERIAL VEHICLES   
  
 Ariel Perez-Rosado, Doctor of Philosophy, 2016 
  
Dissertation directed by: Hugh A. Bruck, Professor, Department of 
Mechanical Engineering 
 
 
Flapping Wing Aerial Vehicles (FWAVs) have the capability to combine the benefits 
of both fixed wing vehicles and rotary vehicles.  However, flight time is limited due 
to limited on-board energy storage capacity.  For most Unmanned Aerial Vehicle 
(UAV) operators, frequent recharging of the batteries is not ideal due to lack of 
nearby electrical outlets.  This imposes serious limitations on FWAV flights.  The 
approach taken to extend the flight time of UAVs was to integrate photovoltaic solar 
cells onto different structures of the vehicle to harvest and use energy from the sun.  
Integration of the solar cells can greatly improve the energy capacity of an UAV; 
however, this integration does effect the performance of the UAV and especially 
FWAVs.  The integration of solar cells affects the ability of the vehicle to produce the 
aerodynamic forces necessary to maintain flight.  This PhD dissertation characterizes 
the effects of solar cell integration on the performance of a FWAV.  Robo Raven, a 
  
recently developed FWAV, is used as the platform for this work.  An additive 
manufacturing technique was developed to integrate photovoltaic solar cells into the 
wing and tail structures of the vehicle. An approach to characterizing the effects of 
solar cell integration to the wings, tail, and body of the UAV is also described.  This 
approach includes measurement of aerodynamic forces generated by the vehicle and 
measurements of the wing shape during the flapping cycle using Digital Image 
Correlation.  Various changes to wing, body, and tail design are investigated and 
changes in performance for each design are measured.  The electrical performance 
from the solar cells is also characterized. A new multifunctional performance model 
was formulated that describes how integration of solar cells influences the flight 
performance.  Aerodynamic models were developed to describe effects of solar cell 
integration force production and performance of the FWAV.   Thus, performance 
changes can be predicted depending on changes in design.  Sensing capabilities of the 
solar cells were also discovered and correlated to the deformation of the wing.  This 
demonstrated that the solar cells were capable of: (1) Lightweight and flexible 
structure to generate aerodynamic forces, (2) Energy harvesting to extend operational 
time and autonomy, (3) Sensing of an aerodynamic force associated with wing 
deformation. Finally, different flexible photovoltaic materials with higher efficiencies 
are investigated, which enable the multifunctional wings to provide enough solar 
power to keep the FWAV aloft without batteries as long as there is enough sunlight to 
power the vehicle.   
 
 
  
 
 
 
 
 
 
 
 
 
 
 
DESIGN, FABRICATION, AND PERFORMANCE CHARACTERIZATION OF 
MULTIFUNCTIONAL STRUCTURES TO HARVEST SOLAR ENERGY FOR 
FLAPPING WING AERIAL VEHICLES   
 
 
 
by 
 
 
Ariel Perez-Rosado 
 
 
 
 
 
Dissertation submitted to the Faculty of the Graduate School of the  
University of Maryland, College Park, in partial fulfillment 
of the requirements for the degree of 
Doctor of Philosophy 
2016 
 
 
 
 
 
 
Advisory Committee: 
Professor Hugh A. Bruck, Chair 
Professor Satyandra K. Gupta 
Professor Abhijit Dasgupta 
Professor Sarah Bergbreiter 
Professor Sung W. Lee (Dean’s Representative) 
 
  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
© Copyright by 
Ariel Perez-Rosado 
2016 
 
 
 
 
 
 
 
 
 
 
 
 
  ii 
 
Dedication 
I dedicate this dissertation to my parents Mandy and Ariel and my sister Mariely.  
They have always and will always be my biggest supporters and I will always love 
them for that. 
  iii 
 
Acknowledgements 
 I would like to express gratitude toward my research advisor, Dr. Hugh Bruck, 
for his guidance throughout the entire course of this dissertation.  I learned so much 
about research and life along the way and will always be grateful for the experiences 
and lessons along the way.  I would also like to express gratitude toward Dr. 
Satyandra Gupta for his guidance and help through the years as well.   
 Additionally, I would like to thank Dr. Sarah Bergbreiter, Dr. Abhijit 
Dasgupta, and Dr. Sung Lee for serving on this committee. 
 Looking back at the progress that has been made throughout the years, there is 
no way that this much progress would have been made without the help of a few 
people.  First I would like to thank the other PhD students in the Robo Raven team.  
John Gerdes, Luke Roberts, and Alex Holness have all helped in the development of 
Robo Raven.  
 Lastly, I would like to thank Dr. Byung-Lip “Les” Lee and the ASFOR for 
their financial support. This work was sponsored by AFOSR grant FA9550-12-10158. 
  iv 
 
Table of Contents 
 
 
Dedication ..................................................................................................................... ii 
Acknowledgements ...................................................................................................... iii 
Table of Contents ......................................................................................................... iv 
List of Tables .............................................................................................................. vii 
List of Figures ............................................................................................................ viii 
Chapter 1: Introduction ................................................................................................. 1 
1.1 Background ....................................................................................................1 
1.2 Motivation for Research ................................................................................2 
   1.3    Goal and Scope...............................................................................................4 
Chapter 2: Literature Review.........................................................................................6 
 2.1.1 Vertical Based Flight.............................................................................7 
 2.1.2  Lateral Based Flight............................................................................12 
2.1.3 Summary of Findings in FWAV Design.............................................21 
 2.2       Photovoltaic Solar Cell Technology.............................................................21 
2.3 Multifunctional Design.................................................................................24 
2.4 Force Measurement Techniques...................................................................27 
2.5 Digital Image Correlation Techniques..........................................................30 
2.6 Efforts in Modeling.......................................................................................32 
2.7 Summary of Literature Review.....................................................................33 
Chapter 3: Design and Manufacturing of Robo Raven................................................35 
3.1 Introduction...................................................................................................35 
3.2 System Overview..........................................................................................36 
3.2.1 Design Requirements...........................................................................36 
3.2.2 System Decomposition........................................................................37 
3.3 Design and Fabrication of Wings..................................................................38 
3.3.1 Wing Design........................................................................................38 
3.3.2 Experimental Wing Design Procedure................................................39 
3.3.3 Characterization of Selected Wing Design.........................................41 
3.4 Design and Fabrication of Wing Drive Subsystem......................................43 
3.5 Design and Fabrication of Steering Subsystem............................................46 
3.6 Design and Evaluation of Normal Flapping Gaits........................................47 
3.7 Conclusions...................................................................................................49 
Chapter 4: Design, Fabrication, and Characterization of Multifunctional Wings to     
Harvest Solar Energy in Flapping Wing Aerial Vehicles............................................51 
4.1 Introduction..................................................................................................52 
4.2   Design and Layered Manufacturing Process for Compliant Multifunctional   
Wings.........................................................................................................53 
4.2.1 Design of Compliant Multifunctional Wing.......................................53 
4.2.2     Multifunctional Wing Fabrication......................................................55 
       4.2.3     Integrating Multifunctional Wings into Robo Raven for Energy 
   Harvesting...........................................................................................58 
4.3  Experimental Characterization of Wing Mechanics...................................60 
  v 
 
4.3.1 Measurement of Lift and Residual Thrust Forces...............................60 
4.3.2 3D Digital Image Crrelation (DIC) Characterization of Wing Shape.64 
4.4 Multifunctional Performance Modeling of Wings.......................................68 
4.5  Experimental Results for Multifunctional Performance.............................72 
4.5.1 Direct Powering of Motors...................................................................73 
4.5.2 Recharging of Batteries........................................................................76 
      4.5.3    Impact of Aerodynamic Forces of Power Output Solar Cells for  
 Sensing..................................................................................................77 
4.6 Conclusions..................................................................................................81 
Chapter 5: Integrating Solar Cells into Flapping Wing Air Vehicles for Enhanced   
Flight Endurance..........................................................................................................85 
5.1 Introduction..................................................................................................85 
5.2       Increasing Solar Cell Coverage by Increasing Wing Area..........................88 
5.2.1 Designing and Building Larger Solar Wings.......................................88 
      5.2.2    Modeling Performance Changes Based on Design Changes to the   
Wing.....................................................................................................92 
5.2.3 Wing Testing........................................................................................95 
5.3  Increasing Solar Cell Coverage by Utilizing the Tail...............................101 
5.3.1  Modeling Performance Based on Design Changes to the Tail..........101 
5.3.2 Designing and Fabricating Solar Tails...............................................104 
5.3.3  Tail Testing........................................................................................105 
5.4 Increasing Solar Cell Coverage by Utilizing Body...................................110 
5.5 Results and Discussion..............................................................................111 
5.5.1 Flight Tests.........................................................................................111 
5.5.2  Battery Recharge Tests......................................................................115 
5.6  Conclusions..............................................................................................117 
Chapter 6:  Designing a Flapping Wing Air Vehicle Capable of Continuous Solar 
Flight by Integrating High Efficiency Photovoltaic Solar Cells................................120 
6.1 Introduction...............................................................................................120 
6.2 Solar Cell Technologies............................................................................123 
6.2.1  Amorphous Silicon Solar Cells.........................................................123 
6.2.2  Semi-flexible Polycrystalline Solar Cells.........................................125 
6.2.3  Gallium Arsenide Solar Cells...........................................................126 
6.3  A-Si and Polycrystalline Powered Vehicle..............................................129 
     6.4      Modeling of the Effects of Wing Characteristics on Flapping Power,  
Thrust, and Lift........................................................................................132 
6.5  Wing Design for Integrated GaAs Solar Cells.........................................135 
6.6  Performance of Mock GaAs Wings.........................................................137 
6.7    Expected Electrical Impact of New Design..............................................140 
6.8  Conclusions..............................................................................................142 
Chapter 7: Conclusions..............................................................................................145 
7.1  FWAV Fabrication...................................................................................145 
7.2 Characterization........................................................................................146 
7.2.1 Characterization of Aerodynamic Forces..........................................146 
7.2.2 Characterization of Deformation using DIC.....................................146 
7.2.3 Electrical Performance Characterization...........................................147 
  vi 
 
7.2.4 Characterization of Self Sensing.......................................................148 
7.3  Modeling..................................................................................................148 
7.3.1 Multifunctional Modleing..................................................................148 
7.3.2 Aerodynamic Modeling.....................................................................149 
7.4 Future Work..............................................................................................150 
Bibliography ........................................................................................................... 1522 
  vii 
 
List of Tables 
 
Table 3.1: Properties of the Common Raven, Corvus Corax......................................36 
Table 3.2: Weight breakdown of Robo Raven for electronics, frame, wings, tail, and 
total..............................................................................................................................38 
Table 3.3: Wing Design Parameters............................................................................42 
Table 3.4: Performance Test Results for Futaba S9352HV........................................45 
Table 3.5: Flight test results of the Robo Raven.........................................................49 
Table 4.1: Parameters for multifunctional wing design...............................................54 
Table 4.2: Properties of Robo Raven...........................................................................55 
Table 4.3: Lift and residual thrust loads generated by each wing design....................63 
Table 4.4: Weight and payload characteristics for each UAV design.........................63 
Table 4.5: Correlation coefficients between thrust and lift forces and shear and biaxial 
strains respectively.......................................................................................................68 
Table 4.6: Comparison of predicted and measured flight time for regular and 
multifunctional wings (dnf denotes “did not fly”).......................................................76 
Table 4.7: Comparison of theoretical charging to actual charging results..................77 
Table 4.8: Correlation factors between thrust and % change in power output for each 
wing design..................................................................................................................81 
Table 5.1: Characteristics of the different wing designs..............................................90 
Table 5.2: Strouhal number for each wing design under each flapping condition....100 
Table 5.3: Specifications for solar cell tail designs...................................................104 
Table 5.4: Payload calculation from total force generation.......................................109 
Table 5.5: Flight Characteristics of Robo Raven and Robo Raven IIIv4..................114 
Table 5.6: Flight Time Results for Different Numbers of Solar Cell Modules.........114 
Table 6.1: Comparison of photovoltaic materials......................................................128 
Table 6.2:  Flight Characteristics of Robo Raven III v4 and v5................................139 
Table 7.1: Summary of each version of Robo Raven III fabricated..........................150 
 
 
 
 
 
 
 
 
 
 
 
 
 
  viii 
 
List of Figures 
Figure 1.1: Left) Insect based UAV design [5] Right) UMD avian inspired UAV 
design [1].......................................................................................................................2 
Figure 2.1:  Robotic fly design with a pair of independently actuated wings [5].........9 
Figure 2.2: Laterally controlled movement of robotic fly [5].......................................9 
Figure 2.3: Tethered MAV with rotary actuator [6]....................................................11 
Figure 2.4: SF-3 Mentor Robot [7]..............................................................................11 
Figure 2.5: Nano Hummingbird [8].............................................................................12 
Figure 2.6: Microbat [9]...............................................................................................14 
Figure 2.7: DelFly Micro next to a Euro coin [10]......................................................14 
Figure 2.8: Small Bird [11]..........................................................................................16 
Figure 2.9: Big Bird [11].............................................................................................16 
Figure 2.10: NPS Flapping-Wing MAV [12]..............................................................17 
Figure 2.11: Great Flight Diagram [29].......................................................................18 
Figure 2.12: Flight Cost of Real Birds [53].................................................................19 
Figure 2.13: Wing Characteristics of Actual Birds[54]...............................................20 
Figure 2.14: Calculated Power Needed vs Solar Flux.................................................20 
Figure 2.15: Fixed wing vehicle with vibrational harvesting capability as well as solar 
harvesting capability [13]............................................................................................24 
Figure 2.16: UAV with batteries integrated into the wing [19, 20].............................25 
Figure 2.17: Sensing results showing a correlation between thrust (grey) and strain 
(black) on a FWUAV Top) Time Resolved, Bottom) Angular Position of the Wing 
[23]...............................................................................................................................26 
Figure 2.18:  Load cell test configurations for early UMD UAVs Top) Lift Bottom) 
Thrust [25]...................................................................................................................28 
Figure 2.19: a) Thrust response from load cell b) Lift response from load cell [25]..29 
Figure 2.20: Speckled wing used for 3D DIC [28]......................................................31 
Figure 2.21: Wing contours while flapping [28].........................................................31 
Figure 3.1: Functional Decomposition of Robo Raven: dotted lines (- - -) denote 
energy flows and solid lines (—) denote signal flows.................................................37 
Figure 3.2: Characteristic wing design: S is the semi-span, C is the chord, and tn are 
the diameters of carbon fiber stiffening rods...............................................................39 
Figure 3.3: Custom built test stand fixture with 6 DOF load cell in the wind tunnel..41 
Figure 3.4: Completed wing with geometry defined by Table 3.3..............................43 
Figure 3.5: Lift and thrust production during multiple flapping cycles.......................44 
Figure 3.6: Main drive assembly (exploded view)......................................................46 
Figure 3.7: Tail drive subsystem assembled (left) and exploded view (right).............47 
Figure 3.8:  Fully Assembled Robo Raven..................................................................48 
Figure 4.1: (left) Robo Raven III, the first solar powered robotic bird using 
multifunctional wings, and (right) parameters for the multifunctional wing design: S 
is the semi-span, C is the chord, and   are the diameters of carbon fiber stiffening 
rods...............................................................................................................................54 
Figure 4.2: Layered manufacturing process for multifunctional solar cells wings 
(bottom side of wing is shown)....................................................................................56 
  ix 
 
Figure 4.3: (Left) Assembled multifunctional wing with 6 solar cell modules, (Right) 
multifunctional wings integrated into Robo Raven III................................................57 
Figure 4.4: (Left) original Robo Raven III wing design with 11 solar cell modules, 
and (Right) the modified wing design.........................................................................58 
Figure 4.5: (Left) Schematic of battery charging circuit used for multifunctional wing 
structures, and (Right) the actual wiring of the solar cells for the circuit....................59 
Figure 4.6: Test stand used in wind tunnel to characterize aerodynamic lift and thrust 
loads on the compliant multifunctional wings.............................................................61 
Figure 4.7: Time resolved load cell results for all four wing designs: (top left) regular, 
(top right) 6 module, (bottom left) 11 module, (bottom right) modified 11 module...61 
Figure 4.8: Comparison of out-of-plane displacement (W) for each wing at the 
horizontal position while flapping downward, which was found to be most 
representative of the relative deformations between the wing designs for the 20 
different wing positions that were measured during a single flapping cycle. (top left) 
regular, (top right) 6 module, (bottom left) 11 module, (bottom right) modified 11 
module. Dashed lines indicate approximate location of the modules..........................65 
Figure 4.9: Comparison of time resolved residual thrust and shear strain: (top left) 
regular, (top right) 6 module, (bottom left) 11 module, (bottom right) modified 11 
module..........................................................................................................................66 
Figure 4.10: Comparison of time resolved aerodynamic lift and biaxial strain relative 
to mean biaxial strain for the entire wing: (top left) regular, (top right) 6 module, 
(bottom left) 11 module, (bottom right) modified 11 module.....................................67 
Figure 4.11: Recharging time profiles for the 12 and 22 module UAVs.....................77 
Figure 4.12: (a) Electrical power generated by the 12 and 22 module UAVs while 
flapping, and thrust versus % change in power output for (b) 22 module and (c) 12 
module UAVs indicating that the solar cells can be used to sense aerodynamic forces 
due to the shear deformations......................................................................................80 
Figure 4.13: Shear strain versus % change in power output for (a) 22 module and (b) 
12 module UAVs.........................................................................................................81 
Figure 5.1: Robo Raven I [38].....................................................................................86 
Figure 5.2: Parameters for wing design: S is the semi-span, C is the chord, and tn are 
the diameters of carbon fiber stiffening rods...............................................................88 
Figure 5.3: Wings designed, built, and characterized to determine the effects of solar 
cell integration on flight endurance for Robo Raven...................................................91 
Figure 5.4: Test stand with residual thrust and lift directions identified. The ATI 
Mini40 6 DOF load cell is capable of measuring up to 40 N of force with a resolution 
of 0.01 N in the thrust direction and 120 N of force with a resolution of 0.02 N in the 
lift direction..................................................................................................................96 
Figure 5.5: Load cell results and subsequent changes in payload capacities compared 
to predicted results from models for different wing designs: Top) Residual Thrust, 
Middle) Aerodynamic Lift, Bottom) Difference in Payload. Mass of each single wing: 
Wing A = 19.3g, Wing AS = 45.2g, Wing B = 25.4g, Wing BS = 50.0g...................99 
Figure 5.6: Solar cell tail designs that were tested: a) Tail 1 b) Tail 2 c) Tail 3 d) Tail 
4 (dimensions in cm)..................................................................................................105 
Figure 5.7: Comparison of the drag from each tail design while wings are stationary 
and vehicle at a 0° angle of attack.............................................................................106 
  x 
 
Figure 5.8: Comparison of the residual thrust from each tail design while wings are 
flapping and vehicle pitched at 20°............................................................................108 
Figure 5.9: Comparison of the lift from each Tail Design while wings are flapping 
and vehicle is pitched at 20 degrees. Mass of each tail: Tail 0 = 6.2g, Tail 1 = 16.0g, 
Tail 2 = 13.1g, Tail 3 = 17.0g, Tail 4 = 14.0g...........................................................109 
Figure 5.10: Two different design configuration for integrating 6 additional solar cell 
modules: (Top) 2 modules in body, 4 modules in tail; (Bottom) 3 modules in body, 3 
modules in tail............................................................................................................111 
Figure 5.11: Robo Raven IIIv4..................................................................................113 
Figure 5.12: Recharging circuit that regulates voltage going to the battery for safe 
recharging where 4.2V are supplied to the battery at the blue node and 8.4V are 
supplied at the red node.............................................................................................116 
Figure 5.13. Comparison of recharging results..........................................................117 
Figure 6.1: Previous Robo Raven & Robo Raven III design: Top left: Robo Raven I, 
Top right: Robo Raven III v1, Bottom left: Robo Raven III v2, Bottom right: Robo 
Raven III v3...............................................................................................................122 
Figure 6.2: Powerfilm MPT6-75 A-Si Solar Module................................................124 
Figure 6.3: Polycrystalline Solar Cell........................................................................126 
Figure 6.4: Sample GaAs photovoltaic solar module................................................127 
Figure 6.5: Polycrystalline tail and body integrated to Robo Raven III v4...............130 
Figure 6.6: Robo Raven with polycrystalline solar cells in the tail and body, and 
Powerfilm flexible solar cells in the wings................................................................131 
Figure 6.7: Left: Mock GaAs wings Right: Completed Robo Raven III v5 design..137 
Figure 6.8:  Time resolved results of Robo Raven III v4 and v5...............................138 
Figure 6.9: Average lift and thrust generated and compared to the predicted 
results.........................................................................................................................139 
Figure 6.10: Potential Power produced by Robo Raven III v5 wing design.............141 
Figure 6.11: Power produced by available solar cell technology for Robo Raven III v5 
wing design................................................................................................................141 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  xi 
 
 
 
  1 
 
Chapter 1: Introduction 
1.1 Background 
 Unmanned Aerial Vehicles (UAVs) have gained a significant amount of 
attention in recent years due to their versatility in a wide range of missions and 
applications.  UAVs are heavily utilized in the military for reconnaissance missions.  
With onboard cameras and sensors, specialists can observe an area of interest without 
endangering any lives.  Similarly, UAVs can also be used in rescue efforts where an 
inhospitable area can be surveyed quickly.  UAVs have also gained considerable 
attention from civilians that have found several commercial applications for these 
vehicles.  Farmers could easily use a small UAV to survey their fields and check on 
their crops saving time, fuel, and energy. UAVs have even been used for 
cinematography and video production to safely obtain an aerial shot for certain 
scenes.  Besides the aforementioned military and commercial uses, hobbyist have had 
a huge interest in UAVs for years.  There are model airplanes, helicopters, and 
quadrotors that have been built by people from all walks of life.   
 UAVs can be broadly classified into rotorcraft, airplanes, or flapping wing 
aerial vehicles (FWAV).  Rotorcraft are very maneuverable and generate a large 
amount of lift.  This enables them to hover very well, but lack the speed to cover a 
large distance in a short amount of time.  On the other hand, airplanes or fixed wing 
aircraft use thrust force to generate aerodynamic lift.  This means they can cover large 
distances very quickly, but cannot hover and lack the maneuverability of a rotorcraft.  
FWAVs are unique in that they provide a versatile compromise between these two 
  2 
 
extremes.  Depending on how they are designed, FWAVs can hover just like 
rotorcraft or cover large distances like fixed wing aircraft.  Smaller scale FWAVs 
take inspiration form insect based flight [5].  These FWAVs can hover and move very 
much like rotorcraft.  Larger scale FWAVs take more inspiration from avian flight 
[1].  These UAVs can achieve greater speeds than their smaller counterparts while 
still maintaining more maneuverability than fixed winged aircraft. 
 Figure 1.1: Left) Insect based UAV design [5] Right) UMD avian inspired UAV 
design [1] 
 
 Prior work at the Advanced Manufacturing Laboratory (AML) in the 
University of Maryland has demonstrated the utility of FWAVs as research platforms 
and as practical flying prototypes capable of transmitting live video, morphing wings, 
and having relatively large payload capacities [1]. These UAVs used a single DC 
motor to drive the wings in a synchronized symmetric manner.  The FWAVs 
developed previously in AML lay the groundwork for the work presented in this 
dissertation. 
1.2 Motivation for Research 
 A major factor in the performance of an UAV is onboard energy capacity.  
For UAVs that are powered and flown with an onboard battery, the size of the battery 
is limited by the payload capacity of the vehicle.  This limitation hinders the flight 
  3 
 
endurance for the UAV, a problem seen throughout the UAV field.  A possible way 
to overcome this challenge would be to increase the available energy capacity by 
using on-board solar cells. This allows for increased flight time while decreasing the 
payload contribution of a large power source, thus potentially allowing for either: (1) 
battery size reduction or battery removal with the same performance capability, or (2) 
an increase in overall payload capacity.  By integrating solar cells into the existing 
structures of UAVs, multifunctional structures are created.  Multi-functional 
structures combine multiple functional requirements into a single structural 
component to create better efficiency in the overall design.  Successful development 
of multifunctional structures can be expanded to other UAVs, including fixed wing 
and rotary craft.  However to integrate commercial solar cells to the structures of an 
UAV, a large UAV must be developed to have enough surface area for solar cell 
integration.   
 It is unclear how the integration of photovoltaic (PV) solar cells will effect 
vehicle performance.  Integrating these solar cells will add mass to the vehicle, 
causing inertial changes to various structures of the vehicle.  Also, integrating PV 
solar cells into the wing will alter wing deformation. PV solar cells are expected to 
alter the stiffness of any vehicle component that they are integrated into. This change 
in stiffness alters how the wings deform while flapping. Wing deformation is an 
important aspect in force generation [1].  Altering the wing design is expected to 
affect wing force generation.  However, due to a lack of wing models for FWAVs, we 
cannot predict the severity of these changes.  It is clear that solar cell integration can 
be beneficial by increasing the vehicle’s onboard energy capacity; however, how 
  4 
 
beneficial and at what cost are unknown.  This dissertation aims to determine the 
benefits of PV cell integration and the cost in performance associated with that 
integration.   
 
1.3 Goal and Scope 
 
 This work aims to characterize solar cell integration into the structures of a 
flapping wing unmanned aerial vehicle.  Successful integration can lead to self-
charging and increased flight endurance.  With efficient enough solar cells, infinite 
flight time with direct sunlight may be achieved.  However, it is the modeling work in 
multifunctional UAV design that has the most significant impact on the field. The 
goals of this work are as follows: 
 
1. Develop an approach to integrate photovoltaic solar cells to the existing wing 
structure of the UAV: The wings offer the largest surface area for solar cell 
integration.  The wings are also responsible for the lift and thrust generation of a 
FWAV.  A new manufacturing technique is required to integrate these cells while still 
maintaining the necessary compliance for flight.   
 
2. Characterize the impact of solar cell integration: New wing designs with 
solar cells integrated into the wings must be developed, characterized, and tested.  
Changes in wing deformation must be measured and characterized to understand how 
these design changes affect performance.  A multifunctional performance model must 
be developed to evaluate and predict how the changes in performance and mass affect 
  5 
 
flight time of the FWAV. Aerodynamic models must be developed to explain how the 
changes in design are expected to alter the performance of the vehicle. 
 
3. Limitations created by solar cell integration: Since the existing wings have 
been completely saturated with solar cells, new larger wings that can incorporate 
more PV cells can be designed.  The changes in performance caused by scaling up to 
larger wings must be measured, understood, and modeled.  The tail and body are the 
last two places on the UAV where solar cells can be integrated. The body is not 
responsible for force generation so is only expected to contribute to an increase in 
mass. Since the tail is responsible for control of the UAV, performance is expected to 
be affected.  This change in performance should be quantified and modeled as well as 
the electrical benefits from the additional PV solar cells.   
 
4. Investigate new photovoltaic cell technologies: Newer more efficient PV cell 
technologies exist and must be investigated.  The best technology that will have the 
smallest effect on performance while still providing benefits in electrical power must 
be identified.  Then, the actual effects on performance through solar cell integration 
must be identified by designing a FWAV based off of these cells.  The goal is to 
design a FWAV that can be completely powered by the sun and fly as long as there is 
sunlight without a battery. 
 
  6 
 
Chapter 2: Literature Review    
 In this chapter, literature related to the goals of the proposed work is surveyed.  
Due to the multidisciplinary nature of this work, this literature review covers several 
topics that give intellectual insight to the different issues and challenges faced in this 
work.  First, previous UAV designs and accomplishments are observed.  Next, a 
review of photovoltaic solar cell technology is conducted, followed by a review of 
multifunctional design.  Previously used measurement techniques are also reviewed 
as well as modeling efforts from the UAV community. 
 In Section 2.1 previously built UAVs are observed.  In doing so, a clear 
distinction can be made from what has already been accomplished and areas where 
significant progress can be made.  This section serves to explain how the developed 
FWAV is different from what has already been built, and how the FWAV developed 
progresses the field of UAVs. 
 Section 2.2 is a review of photovoltaic (PV) solar cell technology.  Certain 
information on PV solar cells must be known if we plan to integrate them to the 
structures of an UAV.  Here we investigate different PV technologies, the mechanical 
properties of the PV solar cells, and how current UAV designs integrate them to their 
systems.   
 Section 2.3 reviews aspects of multifunctional design.  A general overview of 
multifunctional design is introduced to help define our goal for solar cell integration. 
Current UAVs with multifunctional designs are also observed. 
 Measurement and data collection techniques are very important in obtaining 
information about UAV performance.  This information is also critical for model 
  7 
 
generation.  To compare the results obtained in this work to any previous work, an 
understanding of how the previous data was collected is important. Section 2.4 
investigates current and previous methods being used to measure the aerodynamic 
forces being generated by UAVs.  Section 2.5 investigates how to extract the 
deformation of the wings through Digital Image Correlation (DIC).   
 Section 2.6 covers information that previous researchers have been able to 
determine, through their efforts, about certain aspects that enable these UAVs to 
obtain flight. This section also investigates current modeling efforts for FWAV 
wings.  It is with this section that we aim to identify key components of wing design 
that have a direct correlation to flight performance.   
2.1.1  Vertical Based Flight 
 
 The majority of research thus far has been accomplished on the micro scale.  
These small vehicles are known as Micro Aerial Vehicles (MAVs).  At this size, 
flight behavior mimics the flights of insect and small birds.  Researchers design these 
MAVs to fly vertically and to be capable of great maneuverability. The following 
section investigates the advances in vertical-based flight to understand the challenges 
other researchers have faced. 
 Vijay Kumar designed and created the smallest autonomous quadrotors 
capable of exploring, mapping and scouting an unknown three-dimensional building-
like environment [2]. This UAV is capable of vertical flight with very controllable 
maneuverability (similar to insects and small birds).  Even though this UAV is not a 
FWAV, it runs into similar challenges faced by FWAVs.  Due to the necessary 
onboard components needed to achieve the mapping capabilities, the UAV is limited 
  8 
 
to a minimum length of 0.75m and a mass a little less than 2 kg.  This leads to a 
power consumption of over 400 Watts.  With the best available battery, this UAV is 
only capable of a mission time of 10-20 minutes.  This work introduces the need for 
more power or a more efficient system.  Since this is a quadrotor design, the only way 
to obtain more power is through advances in battery technology. 
 The MFI project at the University of Berkeley [4] is looking at using flapping 
wings instead of rotor technology.  They are studying wing behavior with 2 degrees 
of freedom.  Their design calls for a wing with a length of 25cm with an average 
chord length of 6.7cm.  Though these wings are not small, they are 2.3 mm thick 
which make them more comparable to the behavior of wings used for MAVs.  The 
wings experience a 90 degree rotation at the end of each up and down stroke with a 
140 degree flapping range.  This group is simply observing flapping wing behavior 
and do not have an actual flying vehicle; however, they are collecting the 
aerodynamic forces generated by the flapping wing.  To do so, they use a 2 degree of 
freedom (DOF) load transducer attached to the supporting structure of the wing.  
Minimizing friction, they are able to collect the lift and thrust forces generated by the 
wings.  The flapping velocity and rotation of the wing are altered and the results are 
compared.  Using their methods of data collection they are able to decouple the 
rotational contributions of the wings from the simple flapping motion. 
 The wings used in the MFI group are very similar in design to the wings used 
in Robert Wood’s group out of Harvard [5].  They created a MAV at a very small 
scale.  They were able to achieve much more flight control by using two piezoelectric 
actuators. 
  9 
 
 
Figure 2.1:  Robotic fly design with a pair of independently actuated wings [5] 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 2.2: Laterally controlled movement of robotic fly [5] 
 
  10 
 
 Having two piezoelectric actuators allows for more flight control.  Flapping 
one wing differently than the other alters the symmetry causing the body to roll and 
change the trajectory of the vehicle. Independent wing control is certainly an 
advantage to gaining maneuverability and is something that can be applied to larger 
more lateral based FWAV designs.  However, a major drawback of these vehicles is 
their power source.  Because they are so small and demand such high power, these 
vehicles are tethered to a nearby power supply.  This can be seen in Figure 2.1, where 
the two copper wires are attached to the moving vehicle.  This is a huge limitation to 
this design.  A battery may simply add too much payload at this scale.  Even with a 
battery small enough, it may not provide enough power to achieve the same flight 
observed experimentally.   
 Another small tethered design from Virginia Tech uses a rotary actuator [6]. 
The flight dynamics are very similar to the Berkeley and Harvard MAVs however 
this MAV uses a small pager motor to drive the wing instead of the piezoelectric 
actuator.  The use of a motor to drive the wings makes this design easier to scale up.  
This design also introduces a tail consisting of 4 flaps to stabilize the MAV’s flight.  
Instead of using independent wing control to alter the vehicle’s flight path, the tail is 
actuated to control the vehicle’s flight.  This MAV is tethered to a power supply, but 
it can be easily scaled to a larger body using DC motors and available batteries.  This 
was the direction taken with the SF-3 Mentor robot [7].  This vertical flight UAV is 
much larger than the FWAVs previously mentioned.  It weighs 440 grams with a 0.36 
meter wingspan.  Unlike the previous designs, this UAV uses 4 wings, and is 
powered by a NiCd battery pack. However, the power required to lift 440 grams was 
  11 
 
very demanding on the 8 pack NiCd battery used and as a result the maximum flight 
time of this UAV was only 20 seconds.  However short the flight may have been, it 
was able to fly without a tether. 
 
Figure 2.3: Tethered MAV with rotary actuator [6] 
 
 
Figure 2.4: SF-3 Mentor Robot [7] 
 
 The final vertical flight MAV observed was the Nano Humingbird [8]. This 
vehicle is similar to the previous two in that it is powered by a DC motor, but unlike 
  12 
 
the Mentor Robot it is much smaller scale. It weighs a total of 19 grams and has a 
wingspan of 16.5 cm.  With less weight and a smaller battery, the maximum flight 
time was 4 minutes.  What is interesting about this system is that the linkage that 
transfers power to the wings is cable driven.  Turning is done by changing the tension 
at the bottom of the wings.  This shows that by tensioning one wing more than the 
other, the performance of the FWAV is altered.  Powering this much smaller design is 
a Lihium Polymer battery.  The result is a tailless hummingbird-like robot. 
 
Figure 2.5: Nano Hummingbird [8] 
 
2.1.2 Lateral Based Flight 
 
  When transitioning from vertical flight to lateral flight, there are a few 
common consistencies.  First, the size of the UAV tends to increase.  This is in 
agreement with flight dynamics observed in nature. Smaller animals, such as insects 
  13 
 
and small birds, have the ability to hover and use a high flapping frequency.  Larger 
birds propel themselves forward and have a more lateral flight path.  They also flap 
slower than smaller animals.  For these vehicles, the body of the MAV is no longer is 
seen as an upright object but as a mostly horizontal object.  This is conveyed in flight 
and how people perceive these vehicles.  Pictures are no longer taken with the UAV 
standing vertically, but with the body horizontal.  These UAVs are built to travel 
larger distances and therefore are mostly battery powered.  An external power supply 
is simply no longer an option. 
 The Microbat was the first FWAV to use an onboard battery [9].  A Ni-Cd 
battery was used to power a small DC motor with linkage that flaps the wings 
simultaneously.  This palm-sized FWAV weighs 12.5 grams and is capable of a 42 
second flight.  By adjusting the frequency of this MAV the vehicle was able to 
control its pitch, while adjusting the vertical rudder on the tail allowed for left and 
right control.  Two challenges were faced in the design of the Microbat. First, the 
weight constraints were met for the small size of this vehicle.  Going any smaller 
would not allow the vehicle to fly due to the requirements for power production. 
Second, the size and energy capacity of their battery was a huge limitation.  For a 
UAV of this size, the only way to improve energy capacity is through battery 
technology.  A smaller more powerful battery with a higher energy capacity is the 
only way to increase flight time. 
  14 
 
 
Figure 2.6: Microbat [9] 
 
 In the Netherlands, the Aerospace Software and Technologies Institute of the 
Technical University of Delft developed the Delfy [10]. This small UAV consisted of 
a DC motor that was used to drive a cranking mechanism to flap four wings. It is very 
similar in design to the Microbat [9]. 
 
Figure 2.7: DelFly Micro next to a Euro coin [10] 
 
 The Delfly group’s goal was to reduce the size of their platform as much as 
possible.  They discovered is that was easy to reduce the physical size of their 
  15 
 
structural components with today’s manufacturing capabilities.  The difficult part was 
reducing the weight.  Components that they did not have the luxury of manufacturing 
such as the motor and battery could not be scaled down any further.  If they were to 
scale down the manufactured components, the vehicle would not be able to produce 
enough aerodynamic force to lift the larger weight of the motor and battery.  These 
scale-power limitations play a major factor in UAV design. 
 Since 2009, the University of Maryland’s Advanced Manufacturing 
Laboratory has been designing, building, testing, and flying FWAVs [11].  
Previously, the focus of these research platforms was to develop new manufacturing 
techniques for FWAVs, but they do serve as a stepping stone for continuing UMD 
UAV research.  Small Bird and Big Bird are two UAVs that came out of this research 
effort that have very similar designs.  They flap their wings with a single degree of 
freedom and use a horizontally moving tail for maneuverability.  The wings are used 
to produce the aerodynamic forces necessary for lift and thrust while the tail 
movement controls the turning.  The passive deformation of th wings is what allows it 
to produce enough force to achieve flight.  The small bird uses a magnetic actuator as 
the tail driver while the big bird uses a small servo motor.  Big bird weighs more and 
therefore is scaled up accordingly. 
  16 
 
 
Figure 2.8: Small Bird [11] 
 
 
Figure 2.9: Big Bird [11] 
 
 An unconventional FWAV design was generated by the Air Force Institute of 
Technology in their NPS Flapping–Wing MAV [12].  This UAV uses an actuator that 
opens and closes in a clapping motion to move two wings at the rear of the vehicle.  
At the front of the vehicle is a stationary airfoil that uses simple airfoil theory to 
generate lift.  The two rear wings are driven to generate substantial thrust.  This UAV 
is a hybrid between a fixed wing vehicle and a flapping wing vehicle. All forces are 
not reliant on the flapping wing but shared among the flapping wing and airfoil 
structure.  Although this FWAV is vastly different in design compared to the previous 
  17 
 
FWAVs mentioned, its efficiency of this vehicle is remarkable.  The NPS Flapping–
Wing MAV is capable of a 15 minute continuous flight.   
 
Figure 2.10: NPS Flapping-Wing MAV [12] 
 
 Typically, in nature the when comparing wingspans, the smaller wingspans 
are more capable of vertical flight than larger wingspans.  Where insects and 
humming birds can simply take off vertically, larger birds need some forward 
movement to initiate flight. Figures 2.11 and 2.12 demonstrate just that.  Based off of 
the power produced by available solar cell technologies (W/m2) [45], we can begin to 
understand the FWAV that needs to be developed to have enough surface are for solar 
cell integration to power itself. 
 
  18 
 
 
Figure 2.11: Great Flight Diagram [29] 
 
  19 
 
 
Figure 2.12: Flight Cost of Real Birds [53] 
 
 Based off of the avian inspired flight [53], we can understand the mass to 
power ratio of most birds. The mass to wingspan squared for birds can be observed in 
Figure 2.13. The power for scaling appears to follow P = c*Mass(0.5685) where P is the 
power required, and c is a fitting constant (26.83).  Following this relationship and 
knowing the power that can be produced by current solar cell technology, we can 
develop a plot for the power needed vs wingspan.  A small bird with a wingspan less 
than 0.1m simply does not have enough surface area for the solar cells to power the 
FWAV. This means a wingspan of at least 0.1m is needed for solar powered flight.  
This a already in the horiozntal flight regime as depicted in Figures 2.11 and 2.12. 
The wingspan needed comparable to small birds like the Red Knot to birds the size of 
Albatrosses. 
  20 
 
 
Figure 2.13: Wing Characteristics of Actual Birds [54] 
 
 
 
Figure 2.14: Calculated Power Needed vs Solar Flux 
 
 
 
 
  21 
 
2.1.3 Summary of Findings in FWAV Design 
 
 This section reviewed several UAV designs to determine where the FWAV 
field is and where it needs areas of improvement.  Clearly FWAVs have the 
advantage of taking on characteristics of fixed wing UAVs and rotary wing UAVs.  
The vehicles shown varied by design and size. Some were able to hover and some 
were able to cover some distance.  Limitations were also observed in terms of scale 
and power.  Current battery technology only allows these FWAVS to fly for short 
periods of time.  The longest flight time observed was 15 minutes.  There is clearly a 
need for more power, but a larger battery is not ideal since it adds weight to these 
vehicles that already have a limited payload capacity.  The best solution is to harvest 
energy during flight.  Efforts towards this idea have already been implemented in 
other platforms.  In Virginia Tech, energy harvesting techniques have been 
implemented to a fixed wing UAV [13].  This UAV has cantilever vibrational 
harvesting modules on the wings to harvest energy from wing vibrations and small 
photovoltaic panels on each wing. These efforts assist the battery to have an overall 
longer flight time.  Since the wing is fixed, these techniques are much easier to 
implement for their vehicle; however, for a FWAV it is much more difficult.  The 
onboard payload capacity is much smaller and the wing must deform making 
integration of solar cells much more difficult.   
2.2  Photovoltaic Solar Cell Technology 
 
 To effectively integrate solar cells into any UAV, a general knowledge of 
current photovoltaic solar cells must be acquired.  In this section we review the 
different kinds of flexible PV cells available as well as their mechanical properties. 
  22 
 
This review is very important because unlike fixed wing UAVs with PV cells [13]. 
Putting them on the wing of FWAVs can alter their flight characteristics by changing 
how the wing deforms.   
 Shah et al. made the case for flexible solar cells [14].  They showed that 
besides the several new application for thin film solar cell technologies, the amount of 
material and energy needed to continue to create the currently favored crystalline 
silicon solar cells will run out.  Different kinds of solar cell technologies were 
investigated.  Amorphous Silicon (a-Si), Cadmium Telluride (CdTe), and Copper 
Indium Gallium Selenide (CIGS) were the reported candidates.  Shah et al. also 
touched on polycrystalline or multicrystalline silicon solar cells.  These cells tend to 
be less stiff than their monocrystalline counterparts but are less efficient.  These are 
commercially available with an efficiency of 17-18%.  Commercially available solar 
cells of A-Si were reported to be between 4-8% efficient, were some laboratory 
samples reported a record efficiency of 13%.  Commercially available samples of 
CdTe cells were reported to have 16% efficiencies.  Finally, commercially available 
samples of CIGS were found to have 18.8 % efficiency.  However, the market for 
CIGS solar cells and CdTe Solar cells has dried up.  Since this study was performed a 
new thin film photovoltaic solar cell technology has emerged.  Gallium Arsenide 
(GaAs) photovoltaic solar cell have become the newest most efficient solar cell 
technology.  GaAs solar cells are normally 28% efficient [44], surpassing even 
monocrystalline technology.  AltaDevices recently commercialized a 28.8% 
efficiency GaAs cell and MicroLink Devices a 31% efficient CIGS cell.  In terms of 
tin-film solar cell efficiency, GaAs cells are the best choice.  However,  
  23 
 
 In choosing the best PV cell for our application other factors besides 
efficiency must be observed. The mechanical substrate that the PV material is 
deposited on plays a major role in wing deformation and weight.  This eliminates 
CdTe cells since their substrate is limited to rigid glass [15].  Cost is another major 
aspect that determines which type of PV cell we can obtain.  A-Si and polycrystalline 
solar cells are relatively cheap compared to the CIGS or GaAs cells.  This is due to 
the difficulty in the manufacturing process for A-Si cell compared to the CIGS and 
GaAs cells.  When we need a high quantity of solar cells and can afford to not worry 
about efficiency, A-Si PV cells are the best choice. However, to really stretch the 
electrical capability of the system, GaAs PV cells are the best yet much more 
expensive. 
 To integrate these cells into the wings of an UAV, the mechanical properties 
of these cells must be known.  This data serves as the basis for predictions of 
mechanical failure and efficiency loss under more complex loading situations like 
ours.  Antartis et al. investigated the residual stress and mechanical properties in A-Si 
PV thin films [16]. They found that the performance of the PV cells was found to be 
unaffected after 100 loading cycles at up to 0.3% strain. However, degradation 
occurred when the cells were gradually loaded to 1% strain.  It was found that the 
buckling stress of the A-Si cell measured was (-114 ± 27) MPa and the mean residual 
stress was (-661 ± 93) MPa.  This results in a maximum allowable strain of 0.1 ± 
0.02% for the entire cell. In Blakers et al. they use bending equations to determine the 
minimum survivable radius (ρ) for a curved piece of A-Si cells.   
 
  24 
 
 
Where E is the modulus of Elasticity (168 GPa), x is the half thickness of the piece of 
silicon, and σ is the tensile yield strength (7.0 GPa).  Depending on the thickness of 
the cells we obtain, we can safely bend them as along as we do not reach the 
minimum curvature necessary to cause damage to the solar cells. 
2.3 Multifunctional Design  
 
 By integrating PV solar cells into the various functional structures of an UAV, 
we are implementing multifunctional design.  Multifunctional structures combine 
multiple functional requirements into a single structural component to create better 
efficiency in the overall design.  Multifunctional design has already been used in 
biologically inspired technologies for electromagnetic functionality, heating 
functionality, healing functionality, and sensing functionality [17].  Multifunctional 
structures can be used in miniaturization [18], which is an important concept in 
current MAV design.  As we have already seen out of the group from VA Tech [13].  
The wings are also being used to harvest both solar and vibrational energy on a fixed-
wing vehicle. 
 
Figure 2.15: Fixed wing vehicle with vibrational harvesting capability as well as 
solar harvesting capability [13] 
  25 
 
 
 Another multifunctional UAV consists of integrating the actual batteries into 
the wings of a fixed wing UAV [19, 20].  For this vehicle, the batteries were actually 
integrated into the wing and assist the structure in making an airfoil shape to generate 
lift for the aircraft. This UAV was propeller driven. 
 
 
Figure 2.16: UAV with batteries integrated into the wing [19, 20] 
 
 Multifunctional structures have been used in FWAVs as well. In the 
Department of Mechanical & Electromechanical Engineering at Tamkang University, 
researchers are using a polyvinylidene fluoride (PVDF) skin on the wings for self 
sensing [21, 22].  Although this FWAV does not fly, it plays a critical role in 
determining how wing deformation characteristics correlate to force generation.  
Some self-sensing multifunctional work was also done on the MFI platform [4].  
Strain gauges were placed along the spar components of the wings to observe strains 
during the flapping motion.  At the University of Maryland, elastomeric strain gauges 
  26 
 
were painted onto the surface of the wings of a UAV.  A correlation was found 
between the strains observed and thrust production [23]. 
 
Figure 2.17: Sensing results showing a correlation between thrust (grey) and strain 
(black) on a FWUAV Top) Time Resolved, Bottom) Angular Position of the Wing 
[23] 
 
 Multifunctional design has the ability to make current designs more efficient 
by freeing up volumetric space and allowing for more payload.  Integrating more than 
one functionality into existing structures is beneficial, however, one must always 
worry about how the integration will affect overall performance.  If the change in 
performance in minimal, then adding another functionality to an existing structure is 
beneficial.  An acceptable multifunctional design should minimize the effects that any 
additional integrated function has on the existing performance. 
  27 
 
 Oh et al. [52] also used multifunctional structures on a FWAV.  A vertical 
based flight FWAV was developed that had a thin film membrane that had an antenna 
integrated.  The actual flight performance was not observed, but the capabilities of the 
antenna and performance of the antenna were measured. This is simply another 
example of a multifunctional effort for FWAVs. 
2.4  Force Measurement Techniques 
 
 To evaluate how changes in design affect vehicle performance, the lift and 
thrust forces generated by the vehicle should be measured before and after design 
changes are made.  Similar methods of extracting this data can be observed 
throughout the field.  In the case where a 1 degree of freedom (DOF) load cell is 
used, there are two basic ways to extract the forces.  The first method requires only 
aerodynamic lift to be measured [21, 22].  A wind tunnel is set to the exact speed of 
forward velocity of the UAV.  This is accomplished when the average thrust value 
equals zero.  The only force that should be generated by the flapping UAV is 
aerodynamic lift force.  The load cell is set vertically and the aerodynamic lift can be 
recorded.   
 Another way that a 1 DOF load cell can be used to measure aerodynamic lift 
and thrust forces is to run two tests separately.  This is the method used at UMD for 
many years [11, 23, 24, 25, and 26].  The aerodynamic thrust forces were measured 
with the UAVs wings parallel to the wind direction with the UAV pointed towards 
the wind.  A load cell is attached to a linear air bearing and the resulting forces 
generating during operation are measured and recorded. 
  28 
 
 
Figure 2.18:  Load cell test configurations for early UMD UAVs Top) Lift Bottom) 
Thrust [25] 
 
  29 
 
 
Figure 2.19: a) Thrust response from load cell b) Lift response from load cell [25] 
 
 The lift forces are measured similarly.  For lift the orientation of the UAV on 
the test stand is changed so that the load cell can measure aerodynamic lift.  The test 
stand is also placed in the wind tunnel at an alternate angle so that the UAV is still 
facing the direction the wind is coming from.  Again since these are two separate tests 
the aerodynamic lift and thrust forces cannot be measured and recorded 
simultaneously.  However, this method does not require that the wind tunnel be set to 
the exact forward velocity of the UAV. 
 A much more straightforward and easier way to obtain the aerodynamic forces 
generated by a UAV is to use a 2 DOF load cell [9].  This enables both lift and thrust 
forces to be measured and recorded simultaneously.  The last way to extract the most 
amount of information about flight dynamics is to use a 6 DOF force transducer like 
  30 
 
in Wright State University [27].  This enables the researcher to study lift, thrust, yaw, 
and the moments generated by different UAV designs.   
 Being able to quantify the difference in force generation by altering wing, tail 
and body designs will become a paramount aspect of this work.  A consistent and 
accurate test stand and wind tunnel are critical components in getting accurate and 
comparable results.   
2.5  Digital Image Correlation Techniques 
 
 An important aspect that should be quantified to further understand FWAV 
flight dynamics is wing deformation during the flapping cycle.  The wings of most 
FWAV go through passive deformation while flapping.  The passive deformation is 
responsible for generating the forces necessary for flight.  Further understanding of 
wing deformation and wing shape will encourage better wing design and 
development of a wing model that allows performance to be predicted.  Past attempts 
have been used to extract the wing shape during the flapping cycle.  Previously two 
high speed cameras were used to track fifteen points on a wing surface [26].  This is a 
very tedious way to collect the wing shape since it involves someone going through 
and measuring the displacements one by one for each picture.  It is also prone to error 
since an individual is making the measurements.   
 In Wu et al., three dimensional Digital Image Correlation (DIC) was used to 
track the wing deformation through the flapping cycle [28].  3D DIC uses a system of 
at least two cameras to track the surface of a deforming body.   
  31 
 
 
Figure 2.20: Speckled wing used for 3D DIC [28] 
 
 The surface being observed is painted with a speckled pattern to allow the 
cameras to track up to thousands of points throughout the surface.  Pictures of the 
deforming speckled wing were taken simultaneously with 4 cameras.  The images are 
processed using a DIC software.  This software tracks the speckles in relationship to 
each other in 3D space and determines how much the wing has deformed in 3D space.  
From this information strains along the wing surface can also be determined.  
However, in this case, all that was observed were the contours of the wings during the 
flapping cycle. 
 
Figure 2.21: Wing contours while flapping [28] 
 
 DIC is an important tool used in this work to determine how altering wing 
designs changes the shape of the wings during the flapping cycle. Using this data, 
knowledge can be gained that allows a predictive model for wing performance to be 
  32 
 
made based on how the wing is expected to deform.  This shows a better 
understanding of wing dynamics and wing design for FWAVs. 
2.6  Efforts in Modeling 
 
 Pines et al. describes the challenges faced in future MAV/UAV development 
[29].  One major challenge is the lack of models for MAV/UAV flight behavior.  A 
major contribution to the field would be general models that enable UAV designers to 
anticipate how the UAV will behave under flight conditions.  FWAV design is 
different in that wing design has a huge impact on flight performance.  The generic 
airfoil model will not work for these dynamic structures.  However, previous papers 
and findings can lead us to developing a model for FWUAVs. 
 Generally it has been found that different UAV sizes will encounter different 
Reynold’s Numbers [30].  Smaller UAVs will encounter higher Reynold’s number 
requiring them to flap faster.  Larger UAVs will encounter smaller Reynold’s 
numbers allowing them to flap slower yet still obtain flight.  This is reminiscent of 
behavior and patterns in nature.  
 From work done at the Micro level, we know that Leading Edge Vortices 
(LEVs) play a role in enabling small insects to fly [31, 32].  Under general 
aerodynamic theory, small insect with large thoraxes should not be able to fly; 
however, through LEVs on surface of rapidly moving wings, these animals are able to 
fly.  The small insect like MAVs take advantage of this as well. 
 The rotation of the leading spar also plays a role in wing dynamics.  For 
insect-like flight with front wing rotation [3], the ability for the front spar to rotate 
during the flapping cycle correlated well with lift generation.   This enables us to 
  33 
 
extend traditional translation based wing models since rotation plays also plays a role 
in force generation. 
 Wing stiffness also plays a role in flight dynamics [33].  A wing that is too 
stiff will not deform enough to generate lift; whereas, a wing that is not stiff enough 
will not push the air through the wing in a direction to generate thrust. A balance in 
stiffness is critical in flight dynamics.  
 Using PVDF sensors on the wings showed a correlation between wing 
deformation and lift [21, 22].  Using smaller strain gauges on the wing spars 
correlated the strains to thrust measurements [23].  A better understanding of wing 
shape and kinematics can help to model wing behavior through deformation. 
 These findings all come from observations of wing dynamics.  Putting these 
factors together can help derive a model that will enable designers to mathematically 
predict how a wing will perform before it is even built.  Development of such a model 
will lead to a better understanding of FWAV design. 
2.7  Summary of Literature Review 
 
 This literature review demonstrates the need for the proposed work. Section 
2.1 demonstrates the need for more electrical power from these UAVs.  With the 
maximum flight time currently at 15min, and the limitations imposed by the payload 
capacities of these vehicles, solar cell integration is needed for more power 
generation.  Section 2.2 highlights the benefits A-Si and GaAs PV solar cells.  
Section 2.3 introduces multifunctional design and the anticipated effects that solar 
cell integration will have in vehicle performance.  Sections 2.4 and 2.5 investigate 
data collecting techniques using load cells and DIC.  Section 2.6 introduces the need 
  34 
 
for UAV models to help predict how changes in design will effect flight performance.  
Through the integration of solar cells, the need for more power can be met, but the 
cost in overall performance needs to be understood.  This work serves as a trade-off 
analysis for this issue.  Through this investigation, more knowledge towards FWAV 
behavior is gained allowing for models to be derived that enable performance to be 
predicted based off of changes in vehicle design.   
  35 
 
Chapter 3: Design and Manufacturing of Robo Raven 
 
 In this chapter, a new FWAV named Robo Raven is introduced.  This vehicle 
was designed and manufactured in the Advanced Manufacturing Laboratory at the 
University of Maryland.  This is the base platform used for solar cell integration.  It is 
an ideal candidate thanks to a wingspan of over 1 meter that provides a large surface 
area for PV solar cells. The following is an overview of the development of Robo 
Raven and its characteristics.  This content of this chapter is presented similarly to 
what was published in Soft Robotics [38] 
3.1 Introduction 
 
 There was high interest in developing a platform that can be used to 
investigate outdoor aerobatics maneuvers, so inspiration was taken from the Common 
Raven, a relatively large bird capable of aerobatics. Moreover, to provide 
customizable flapping motions, each wing was powered with a separately controlled 
independent drive train, consisting of a motor, gears, position sensor, and feedback 
control loop. The benefit of this approach is to allow completely arbitrary wing 
motions, which provides increased freedom to explore some aggressive and 
interesting maneuvers.  These include asymmetric roll initiation, upflap-downflap 
asymmetry, gliding and soaring dynamics, gust rejection, and blending of tail and 
wing steering modes. 
 In the present effort, many of the same techniques previously developed to 
characterize and understand changes in wing design are applied.  The results 
culminate in a flying prototype vehicle called Robo Raven, a FWAV that overcomes 
the substantial challenge of producing sufficient lift and thrust for sustained flight 
  36 
 
with fully independent wing control.  This newly designed FWAV has a larger wing 
surface area making it a great candidate for solar cell integration. 
3.2  System Overview 
 
3.2.1  Design Requirements 
 
 The primary objective was to build a FWAV that can be used to learn about 
the effects of changing wing kinematics while improving the maneuverability of 
flapping wing air vehicles outdoors. Since nature provides the best examples of flight 
it was sought to replicate avian-based flight dynamics. This effort was inspired by the 
Common Raven, Corvus Corax, with properties listed in Table 3.1. 
Table 3.1: Properties of the Common Raven, Corvus Corax 
Parameter Common 
Raven 
Unit 
Total Mass 0.69-2.00 kg 
Length 0.63 m 
Wingspan 1.00-1.50 m 
Average Chord 0.21 m 
Aspect Ratio 2.77  
Flight Speed 9.80-12.50 m/s 
 
For vehicle design, the following eight requirements were identified: 
1. Flap wings of at least 0.20m2 surface area at 3.5 Hz. This requirement came 
 by observing raven wings and was necessary to enable outdoor flight.  
2. Flap each wing independently with the ability to synchronize wing motions 
 when needed.  This requirement was needed to perform normal flight and do 
 aerobatics. 
3. Have ability to program wing velocity and position as a function of time.  This 
 requirement was needed to optimize normal and aerobatic maneuvers. 
4. Minimize weight to achieve flight and a climb rate of 0.5m/s.  
  37 
 
5. Achieve minimum turning radius of 10m. This requirement was needed to  
 conduct tests in indoor stadiums and fly in outdoor fields with trees. 
6. Remotely control the flight from a distance of 500m.   
7. Land unpowered at glide speed from a height of at least 3m without 
 sustaining structural damage. 
3.2.2 System Decomposition 
 
 The design requirements for Robo Raven assisted in determining how to 
control and power the entire system.  A functional decomposition that demonstrates 
the energy and signal flow through the system was developed.  This allowed for the 
individual components that were going to be needed for the system to be selected. 
 
Figure 3.1: Functional Decomposition of Robo Raven: dotted lines (- - -) denote 
energy flows and solid lines (—) denote signal flows 
 
  38 
 
 Each component that was selected was the lightest component that was 
commercially available.  This would allow for maximum payload capacity when the 
vehicle is completely assembled.  Additionally, each component was chosen for its 
small size.  This allows for easier mounting on the fuselage of the UAV.  A 
breakdown of the components and weights are shown below. 
Table 3.2: Weight breakdown of Robo Raven for electronics, frame, wings, tail, and 
total 
Component Weight 
(g) 
Drive Actuators 
(2 - Futaba S9352HV) 
136.0 
Tail Actuator (Futaba S3114) 7.8 
Controller (Arduino Nano) 6.0 
Wiring 19.8 
LiPo Battery 27.0 
Spektrum 2.4 GHz Receiver 3.0 
Wings (2) 26.0 
Tail 8.0 
Actuator Mount Assembly 
(for the S9352HV servos) 
16.4 
Frame 30.0 
Foam Crash Protection 5.0 
Assembly Fasteners 5.3 
Total 290.3 
 
3.3  Design and Fabrication of Wings 
 
3.3.1  Wing Design 
 
 Wing design was adapted from a previously used approach developed for 
successfully realizing a FWAV platform with synchronized wing motion, which has 
been shown to be effective in generating lift and thrust forces across a variety of 
applications and size scales [11, 23, & 24]. The design is shown in Figure 3.2. 
  39 
 
 
Figure 3.2: Characteristic wing design: S is the semi-span, C is the chord, and tn are 
the diameters of carbon fiber stiffening rods 
 
 The parameters of the wing are as follows: S is the semi-span, C is the chord, 
and tn are the diameters of carbon fiber stiffening rods.  The wing membrane is a 
0.0254mm thick film of biaxially-oriented polyethylene terephthalate (Mylar) which 
provides light weight, flexibility, and good toughness. 
 Due to extensive usage of lightweight materials, this wing design changes 
shape passively in response to loads from the flapping motion.  As the wings flap, 
aerodynamic loads on the compliant wing surface generate a significant camber 
change.  The cambered wing is a thin airfoil, which captures and expels a large 
volume of air with each stroke.  The induced shape change alternates with upstroke 
and downstroke producing lift and thrust. Thus, the selection of parameters in Figure 
3.2 plays a major role generating the aerodynamic loads. 
3.3.2  Experimental Wing Design Procedure 
 
  To identify the best wing design for this new vehicle, a series of experiments 
with different wings were conducted to help characterize this design. The objective of 
this experimental wing characterization was to select a wing geometry that would 
  40 
 
result in the servos operating as close as possible to their peak power operating 
condition. Each wing design was evaluated using a custom-built test stand with a 6 
degree of freedom load cell in a wind tunnel to measure the forces generated by the 
vehicle.  This test stand was mostly made of wood.  It was deigned to keep the 
FWAV at a pitch of 20 degrees, which was the angle taken by the vehicle during 
flight.  An aluminum slab separated the load cell from the rest of the test stand 
providing a smooth flat surface for accurate results.  The load cell used was an ATI 
Mini 40 Force Transducer, capable of measuring a maximum load of 40 N for thrust 
with a resolution of 1/100 N and a maximum load of 120 N for lift with a resolution 
of 1/50 N.  A National Instruments PXI system was used to measure voltages from 
the load cell during wind tunnel testing, and these voltages were post-processed to 
obtain the lift and thrust forces. 
 High speed video footage was recorded during all load cell tests to visually 
observe the difference in flapping for each wing design. These tests were repeated 
under no air flow and significant air flow conditions with varying angles of attack. 
The load cell testing quantitatively revealed which wing created the most lift and 
thrust, while high speed video footage offered qualitative explanations for the results 
and provided design insight.  The test stand is shown in Figure 3.3.  The load cell 
testing quantitatively revealed which wing created the most lift and thrust. 
  41 
 
 
Figure 3.3: Custom built test stand fixture with 6 DOF load cell in the wind tunnel 
 
3.3.3 Characterization of the Selected Wing Design 
 
 Each of the experimentally designed wings were built by hand using a 
template matching the parameters in Table 3.3.  The Mylar membrane was taped flat 
on a clean flat table and carbon fiber tubes were installed according to the template 
using Uhu Por glue. Mylar strips were then used to cover and secure the carbon fiber 
stiffeners on the wing’s membrane, and additional reinforcement Mylar was bonded 
to areas of higher stress concentrations to prevent tearing.  Five holes were melted 
into the wing next to the chord-wise spar and 5 rubber bands were tied through for 
tensioning the wing attachment to the frame.  A completed wing can be seen in Figure 
3.4. 
 
 
 
 
 
  42 
 
Table 3.3: Wing Design Parameters 
Parameter Value Units 
S 605.8 mm 
C 362.0 mm 
t1 3.18 mm 
t2 1.63 mm 
t3 1.63 mm 
t4 1.63 mm 
θ1 20 degrees 
θ2 40 degrees 
 
 
Figure 3.4: Completed wing with geometry defined by Table 3.3 
 
 These wings were tested in a wind tunnel (airspeed approximately 5 m/s) 
using the load cell test stand. The wings generated averages of 243 g of lift and 120 g 
of thrust at a 20 degree angle of attack.  Lift and residual thrust data as a function of 
time from a sample trial are shown in Figure 3.5. 
  43 
 
 
Figure 3.5: Lift and thrust production during multiple flapping cycles 
 
 It is important to note that the response of these wings is similar to the 
response of previously developed FWAVs [25]. The lift demonstrated a smooth 
almost sinusoidal response where the thrust was more complex.  There are several 
peaks within one flapping cycle.  In this case, there are two different magnitudes 
demonstrating the difference in magnitude in thrust produced in the up and 
downstroke at an incline of 20 degrees. 
3.4  Design and Fabrication of Wing Drive Subsystem 
 
 The wing drive subsystem is responsible for producing the wingbeat 
kinematics that result in lift and thrust. The design of this subsystem is an important 
aspect of the vehicle design. Therefore, nature was used for inspiration and provided 
some insight into the trends that exist among feasible designs. 
 Power density is a key design factor for FWAVs to ensure lift and thrust 
forces are sufficient to overcome the weight and drag of the vehicle.  The design of 
custom actuators was beyond the scope of this effort, so commercially available 
  44 
 
actuator for the wing drive subsystem were selected. Vehicle size limited the 
selection to the slower and more powerful actuators, including electric motors, shape 
memory alloys, bimetal benders, and dielectric elastomers.  Of these choices, only 
electric motors and dielectric elastomers offer acceptably high efficiency of 
operation.  However, the very high operating voltage presents integration challenges 
and required additional voltage step-up electronics on-board.  Therefore, electric 
motors were used for this application. 
 Electric motors must be paired with a flapping mechanism to provide 
reasonable speeds of operation.  In addition, integration with a position sensor and a 
feedback loop is required for programmable kinematics.  These requirements lead to 
increased weight, part count, complexity, and integration difficulty.  To avoid the 
challenges associated with matching all of those components and designing the 
required hardware and software, servos were chosen for this effort to actuate the 
wings, due to their high power output, programmable motions, and integrated 
packaging including the motor, drive train, speed controller, and position sensor.  The 
Futaba S9352HV was chosen due to its high figure of merit, computed as the power 
output divided by the mass.  The available servos exhibit a linear relationship 
between available power and figure of merit.  While other servos exist on the market 
that offer either higher torque, higher speed, or lower mass, none are sufficiently 
strong performers in all three categories to justify their selection for our efforts. 
 The Futaba S9352HV servo motors were tested with a dynamometer and 
power analyzer to provide details about the electrical power input as well as the 
mechanical power output. The test results included the power and efficiency curves as 
  45 
 
a function of motor speed.  The objective was to determine the location of the power 
band, the operating condition that provided peak power output, and the operating 
condition that provided peak efficiency. The results of our testing are summarized in 
Table 3.4.  The maximum torque and speed are not reachable simultaneously.  The 
peak power output of 9.4 W is reached at a final drive speed of about 509.9 deg/s 
which corresponds to approximately half the peak torque and speed. 
Table 3.4: Performance Test Results for Futaba S9352HV 
Parameter Value Units 
Stall Torque 12.5 mN-m 
Top Speed 1008.4 deg/s 
Peak Power Output 9.4 Watts 
Peak Efficiency 63.9 % 
Mass 72.0 g 
 
 Due to the importance of weight minimization, 3D printing was used to 
manufacture the frame that comprised the primary structure of the vehicle. The ability 
to produce a part with complex geometry eliminated many assembly steps and 
hardware that would have otherwise been required. The frame was sized to minimize 
the weight and to ensure that it was capable of withstanding the expected load.  The 
part is manufactured using ULTEM 9085 material, which has a tensile strength of 
71.6 MPa.  The design has substantially less stress than this during normal operation.  
Iterative testing has shown that crashes imparting loads of over 5 kg can be survived 
without part failure, yet the weight of the final part is only 8.0 grams. 
 The custom servo horns were laser cut out of 1/4” Delrin®. These servo horns 
fit over aluminum servo horns that physically attach to the servos and have 1/8” holes 
  46 
 
for the wings’ front spars to fit into.  To prevent off-axis loads from damaging the 
gear train inside each servo as well as to provide crash protection, a stiff 3D printed 
nose is attached to the front, labeled Torque Lock in Figure 3.6.  Carbon fiber tubes 
are used to connect the front section to the tail and the rest of the body. An exploded 
view of the CAD model of the main drive system is shown in Figure 3.6. 
 
Figure 3.6: Main drive assembly (exploded view) 
 
3.5  Design and Fabrication of Steering Subsystem 
 
 A steering system consisting of a tail surface mounted to the rear of the 
fuselage was included to control the direction of the FWAV during flight.  The tail 
maintains a fixed elevator angle while providing a variable roll angle relative to the 
fuselage.  A flat trailing edge allows for simplified manufacturing and is also an 
aerodynamically ideal solution.  Any portion of the tail that contracts in span will be 
subject to the upstream wake and produce primarily drag, with a minimal contribution 
to lift.   
  47 
 
 
Figure 3.7: Tail drive subsystem assembled (left) and exploded view (right) 
 
 This design provides the option of fully controlling the steering with the tail 
alone, or blending the steering control with wing motions to explore bio-inspired 
techniques of flight control.  This enables the operator to execute aggressive 
maneuvers.  In addition, this layout gives us the ability to explore some behaviors that 
would be challenging or impossible to train an animal to execute, yet could still be 
informative from a research perspective. Flight testing of this design revealed a 
minimum turn radius of 6 m at a 40 degree tail angle, with an almost linear increase 
in turn radius proportional to reduced tail deflection. 
3.6  Design and Evaluation of Normal Flapping Gaits 
  
 The selection of a flapping gait simultaneously controls the selection of a 
number of important parameters including the flapping amplitude, rate, average 
dihedral, and periodic symmetry, among others.  All of these parameters affect the 
flight speed and body pose, which significantly influence the forces produced during 
flight. At a minimum, our objective with the Robo Raven would be to reach the point 
that corresponds to minimum power required to sustain flight, and thus provides the 
greatest endurance. 
  48 
 
 The prototype vehicle shown in Figure 3.8 was used for flight testing.  
Initially, the angle of attack required adjustment, as early flights exhibited a very high 
forward speed and gradual loss of altitude. This was likely due to the large 
concentration of weight at the front attributed to the servos.  Therefore, a small 
extension was added to the rear of the fuselage from which the battery was 
suspended, which shifted the center of mass rearward, and thus the angle of attack 
increased, directing the net force vector closer to vertical.  The flapping gait was also 
adjusted by increasing the flapping amplitude and slightly increasing the flapping 
rate.  This brought the prototype within the range commonly exhibited by flying 
animals, and resulted in sustained flight with the ability to climb and maneuver. 
 
Figure 3.8:  Fully Assembled Robo Raven 
 
 The flapping gait was programmed on the Arduino Nano. Average force 
production during testing was 242.9 g (5.5 g standard deviation) lift and 119.6 g (6.4 
g standard deviation) thrust.  Due to some turbulence in the wind tunnel a small 
amount of variability was present across trials.  The typical flight duration of the 
  49 
 
system with a 370 mAh onboard LiPo battery is 4 minutes and 30 seconds. Flight 
testing the Robo Raven has helped us to understand the capabilities and operational 
conditions during cruising, which are summarized in Table 3.5. 
Table 3.5: Flight test results of the Robo Raven 
Parameter Value Units 
Flap Rate 4.0 Hz 
Flap Amplitude 100 degrees 
Angle of Attack 20 degrees 
Climb Rate 0.53 m/s 
Reynolds Number 124,000 - 
Strouhal Number 0.395 - 
Minimum Turning radius 6.1 m 
 
3.7  Conclusions 
 
 A new FWAV platform called Robo Raven was developed that flies with 
wings independently controlled by digital servo motors. This represents an 
improvement in flight capability and takes a step closer towards replicating avian 
flight. Thus, this platform represents a significant advance towards biomimicry of 
birds using a FWAV platform.  
 Successfully realizing this design required a multi-faceted approach.  First, 
advanced manufacturing processes such as 3D printing and laser cutting were used to 
create lightweight polymer parts to reduce the weight. Second, wing motion profiles 
were programed that ensured that wings maintain the optimal velocity during the flap 
cycle to achieve the right balance between the lift and the thrust. Third, a method to 
measure aerodynamic forces generated during the flapping cycle was developed. This 
allowed for quick evaluations of many different wing designs to select the best one. 
  50 
 
Finally, a systems approach was adopted to make sure that all components worked 
well as an integrated system. 
 Due to the relatively large size of the FWAV, it is an ideal candidate for solar 
cell integration.  The current maximum flight time is currently 4 minutes and 30 
seconds and the current aerodynamic forces have been measured and recorded.  
Changes in design can be easily tested and quantified using the current methods.   
 
  51 
 
Chapter 4: Design, Fabrication, and Characterization of  
         Multifunctional Wings to Harvest Solar Energy in        
        Flapping Wing Aerial Vehicles 
 
 This chapter is the work presented in Smart Materials and Structures [47]. In 
this section we describe a layered fabrication method that was developed for realizing 
multifunctional composite wings for Robo Raven by creating compliant wing 
structure from flexible solar cells.  The deformed wing shape and aerodynamic 
lift/thrust loads were characterized throughout the flapping cycle to understand wing 
mechanics. A multifunctional performance analysis was developed to understand how 
integration of solar cells into the wings influences flight performance under two 
different operating conditions: (1) directly powering wings to increase operation time, 
and (2) recharging batteries to eliminate need for external charging sources. The 
experimental data is then used in the analysis to identify a performance index for 
assessing benefits of multifunctional compliant wing structures. The resulting 
platform, Robo Raven III, was the first demonstration of a robotic bird that flew using 
energy harvested from solar cells. We developed three different versions of the wing 
design to validate the multifunctional performance analysis. It was also determined 
that residual thrust correlated to shear deformation of the wing induced by torsional 
twist, while biaxial strain related to change in aerodynamic shape correlated to lift. It 
was also found that shear deformation of the solar cells induced changes in power 
output directly correlating to thrust generation associated with torsional deformation. 
Thus, it was determined that multifunctional solar cell wings were capable of three 
functions: (1) lightweight and flexible structure to generate aerodynamic forces, (2) 
  52 
 
energy harvesting to extend operational time and autonomy, and (3) sensing of an 
aerodynamic force associated with wing deformations. 
  
4.1 Introduction 
 
 As a part of previous work, we have developed a highly maneuverable FWAV 
named Robo Raven [38]. These vehicles rely on flapping wings and their deformation 
to generate the aerodynamic forces necessary for flight.  The size of these vehicles are 
comparable to the size of actual birds found in nature.  This platform features 
independently controlled programmable wings. In FWAV such as Robo Raven, flight 
endurance is one of the primary concerns. To perform missions in remote regions, the 
UAV cannot charge batteries using electrical outlets. A possible way to overcome this 
challenge would be to charge batteries using on-board solar cells. Since Robo Raven 
has a large wing area, solar cells can be integrated into the wings.  The resulting 
compliant wing structure with integrated solar cells can be considered both 
multifunctional and smart because it not only provides lift and thrust, but also acts as 
a method of harvesting energy and sensing changes in deformation caused by 
aerodynamic loading, which can be used to determine changes in the flapping profile 
to improve flight control.  This combination allows for increased flight time while 
decreasing the payload contribution of a large power source, thus potentially allowing 
for either: (1) size reduction with the same performance capability, or (2) an increase 
in overall payload capacity.  It also introduces new capabilities for control schemes 
through new sensing capabilities.  Successful development of multifunctional 
compliant wing structures through the integration of multiple functions can be 
expanded to other aspects of all UAVs, including fixed wing and rotary craft. 
  53 
 
 Integrating solar cells into the wings presents the following three challenges. 
First, a new manufacturing process is needed to integrate solar cells into wings 
without substantially increasing weight.  Second, we need to ensure that wings with 
integrated solar cells maintain the appropriate deformation during the flapping cycle 
to ensure production of adequate aerodynamic lift and thrust.  Finally, we need to 
make sure the modified version of Robo Raven with multifunctional wings produces 
enough thrust and lift to compensate for the heavier wings and enable flight. 
 In this chapter, we describe a new layered fabrication method for integrating 
commercial off-the-shelf solar cells into wings for a new solar-powered FWAV: 
Robo Raven III. Different wing designs were tested to observe how adding different 
quantities of solar cells affects flight performance through wing deformation, and 
how this leads to changes in power output that can be measured for potential onboard 
sensing and control.  A new multifunctional performance analysis is also developed 
to quantify the effects of solar cell integration on recharge time and flight time to 
determine trade-offs from the multifunctional effects of solar cell integration to be 
considered by examining the impact of lift and thrust on power requirements versus 
the gains from recharging by harvesting solar energy. 
4.2 Design and Layered Manufacturing Process for Compliant 
Multifunctional Wings  
 
4.2.1  Design of Compliant Multifunctional Wing 
 
 Multifunctional wings were created for Robo Raven III (Figure 4.1). The 
design of the multifunctional wing, also seen in Figure 4.1, was adapted from a 
design we used previously for Robo Raven, which has been shown to be effective in 
generating lift and thrust forces necessary for flight [38]. The parameters of the wing 
  54 
 
are as follows: S is the semi-span, C is the chord, and tn are the diameters of carbon 
fiber stiffening rods. Table 4.1 presents values of the wing parameter used in the 
design reported in this paper. The wing membrane is a 0.001” thick film of biaxially-
oriented polyethylene terephthalate (Mylar) which provides flexibility and toughness 
while remaining lightweight.  Table 4.2 lists the properties of the base Robo Raven 
platform with batteries. 
 
Figure 4.1: (left) Robo Raven III, the first solar powered robotic bird using 
multifunctional wings, and (right) parameters for the multifunctional wing design: S 
is the semi-span, C is the chord, and  are the diameters of carbon fiber stiffening 
rods 
 
Table 4.1: Parameters for multifunctional wing design 
Parameter Value Units 
S 605.8 mm 
C 362.0 mm 
t1 3.18 mm 
t2 1.63 mm 
t3 1.63 mm 
t4 1.63 mm 
θ1 0.358 Rad 
θ2 0.750 Rad 
 
  55 
 
 
 
 
 
Table 4.2: Properties of Robo Raven. 
Parameter Robo 
Raven 
Unit 
Total Mass 0.29 kg 
Length 0.554 m 
Wingspan 1.168 m 
Average Chord 0.248 m 
Aspect Ratio 2.01  
Flight Speed 6.7 m/s 
 
4.2.2  Multifunctional Wing Fabrication 
 
 To produce appropriate aerodynamic lift and thrust forces, many ornithopters 
rely on large deformations using compliant wings at lower flapping frequencies to 
achieve airfoil shapes [24].  The basic compliant wing structure weighs 16.8g with a 
total area of 1420cm2.  To maintain compliance when creating a multifunctional wing 
with a similar structure, Powerfilm’s© MPT6-75 flexible solar cell modules were 
chosen.  These flexible 7.3 x 11.4 cm amorphous silicon solar cell modules are 
reported by the manufacturer to produce 50mA of current at 6V at 100% sunlight 
flux, which represents their maximum power point.  However, the bending stiffness 
and mass of the solar cells as packaged and was much higher than the Mylar, and 
therefore would not allow the wing to deform enough to maintain flight. Therefore, 
modifications had to be made to the solar cells to reduce the mass and bending 
stiffness to be more compatible with the Mylar. By heating and peeling off the 
protective encapsulation on the solar cells, the bending stiffness of the solar cells and 
  56 
 
mass was substantially reduced. Then, solar modules were glued and soldered 
together in parallel to produce more current.  Creating the multifunctional wings from 
the de-encapsulated solar cells modules involved a layered manufacturing process 
(Figure 4.2), and the completed multifunctional wing structure integrated into Robo 
Raven III can be seen in Figure 4.3. 
 
Figure 4.2: Layered manufacturing process for multifunctional solar cells wings 
(bottom side of wing is shown). 
 
 To fabricate the wing, a layered manufacturing process was developed to 
provide precise control over the location of each element of the wing (Figure 2). The 
layered manufacturing process consisted of the following steps:  
(a) 
(f) 
(e) 
(d) 
(c) 
(b) 
  57 
 
(a) A sheet of Mylar is secured to a work table with the use of magnets.  
(b) The wing shape including the hole for the solar modules is cut from the 
 secured sheet of Mylar.  
(c) The solar modules are applied to the wing. 
(d) A Mylar frame is adhered around the solar modules that holds the solar cells 
 in place.  
(e) The spars are held in place using magnetic holders with notches while they are 
 adhered to the wing.  
 
Figure 4.3: (Left) Assembled multifunctional wing with 6 solar cell modules, (Right) 
multifunctional wings integrated into Robo Raven III. 
 
 Once Robo Raven III was completed, a flight test was conducted and it was 
determined that the platform could achieve flight. The first version of Robo Raven III 
used 6 solar cell modules in each wing to generate 300 mA at 6V, so a second row of 
solar cells consisting of 5 modules were used to replace as much of the original wing 
material as possible (Figure 4.4). This version of Robo Raven III was also flight 
tested, and it was determined that the new wing design was incapable of continuous 
flight for more than 10 seconds due to increase in mass and decrease in thrust and lift 
force generation. Based on our previous experience, additional compliance at the 
trailing edge of the wing can compensate for the increase in stiffness over the area of 
  58 
 
the wing when solar modules are integrated, so the wing was redesigned accordingly. 
The modified wing design compared to the original wing design can also be seen in 
Figure 4.4. There are three main differences between the new design and the 
previous.  The first two involve extending carbon fiber tubes in the inside part of the 
wings to permit increase in wing area and compliance of the wing at the trailing edge.  
The final modification involved changing the shape of the Mylar skin into a 
“teardrop”. It was determined that the modified wing design was capable of restoring 
flight capability to Robo Raven III. 
  
Figure 4.4: (Left) original Robo Raven III wing design with 11 solar cell modules, 
and (Right) the modified wing design. 
 
4.2.3 Integrating Multifunctional Wings into Robo Raven for Energy 
Harvesting  
 
 Robo Raven uses a two-cell Lithium Polymer battery rated at 7.4V and 
370mAh. To maintain the balance of the battery cells when charging with the 
multifunctional wings, a charging circuit was design as seen in Figure 4.5. Each 
module produces 50mA at 6V at 100% sunlight flux with the voltage from each wing 
depicted in Figure 4.5 as V1 and V2. A zener diode with a breakdown voltage of 4.3 
V is used to regulate the voltage so it does not exceed the maximum of 4.2 V for each 
cell. The resistor in the circuit was chosen to achieve the appropriate voltage drop 
  59 
 
from the modules based on their current, reducing the power generation by 25%.  The 
drop is transformed to heat and can be felt radiating from the electronics. 
 
Figure 4.5: (Left) Schematic of battery charging circuit used for multifunctional wing 
structures, and (Right) the actual wiring of the solar cells for the circuit. 
 
 For direct powering of the servomotors, the solar cells were directly connected 
to the servomotors instead of the battery. This is possible since servos can operate at 
up to 7.2V and the solar cell output has been measured up to 7.8V. This would 
optimize performance of the solar cells for powering the UAV in series with the 
battery pack, as opposed to the 25% reduction in power experienced when utilizing 
the recharge circuit for the battery. This has an additional benefit of prolonging the 
life of the battery by allowing the solar cells to assist the battery in powering the 
UAV, thereby reducing the current draw on the battery and extending the discharge 
time.   
 
 
 
4.2V Reference 
Wire 
8.4V Reference Wire 
  60 
 
4.3  Experimental Characterization of Wing Mechanics 
 
4.3.1 Measurement of Lift and Residual Thrust Forces  
 
 Since existing computational models are inadequate for accurately predicting 
aerodynamic loads acting on compliant flapping wings, direct measurement of these 
loads during the flapping cycle was selected as the method for gaining insight into the 
effects of wing design parameters on the wing mechanics.  For this study, we adapted 
the previous test stand we developed, but used the same 6 DOF ATI Mini40 load cell 
mounted on a wood and Delrin frame for measuring aerodynamic lift and thrust loads 
simultaneously, as well as the moments generated (Figure 4.6).  The test stand also 
allowed the UAV to be set to any angle of attack from 0 to 20 degrees, which was the 
angle of the bird body relative to the wind direction. Unlike the previous test stand the 
load cell is horizontal to the ground and allows for the lift and thrust forces to be 
measured directly. To simulate the actual flight conditions, the test stand is placed at 
the end of a wind tunnel operating at 6 m/s, which is near the actual flight speed.   
  61 
 
 
Figure 4.6: Test stand used in wind tunnel to characterize aerodynamic lift and thrust 
loads on the compliant multifunctional wings. 
 
 
 
Figure 4.7: Time resolved load cell results for all four wing designs: (top left) 
regular, (top right) 6 module, (bottom left) 11 module, (bottom right) modified 11 
module. 
  62 
 
 
 The wings were testing at a flapping frequency of 4 Hz and were programmed 
to flap with a range of 60 degrees. The wings flapped symmetrically orthogonal to the 
body of the UAV.  Time resolved load profiles for wings with and without solar cells 
can be seen in Figure 4.7 for Robo Raven. These thrust and lift profiles are consistent 
with previous measurements and models of flapping wings where the lift produces a 
sinusoidal profile consistent with aerodynamic drag while the residual thrust exhibits 
a double peak consistent with a “blowback” effect from the rear of the wing during 
the flapping cycle [26, 38]. As a result, the peaks appear 180° out-of-phase when they 
overlap on the time-resolved plot.  
 Comparing these profiles for each wing design, it can be seen that there is a 
slight change in performance caused by the addition of solar cells on the wings.  
Because the solar cells stiffen the wings and reduce compliance in sections of the 
wing structure, it was predicted that the solar cell wings would underperform the 
regular wings.  However, from the profiles it seems that the 6 module wings actually 
have slightly larger values for lift compared to the regular wings.  For the 11 module 
wings, the values for thrust decreased significantly compared to the original wings 
with only a slight increase in lift, which was consistent with the observed loss of 
flight capability.  The modified wing design had an increase in lift force generation 
compared to the original wings, consistent with the observed restoration of flight 
capability.  The average values of lift and residual thrust load for each trial can be 
seen in Table 4.3. 
Table 4.3: Lift and residual thrust loads generated by each wing design 
  63 
 
 
 In a wind tunnel, a residual thrust value of 0 g would correlate to steady-state 
flight conditions. However, our low speed wind tunnel has a maximum velocity of ~5 
m/s, while the actual flight velocity for Robo Raven is 6.7 m/s. Since this meant the 
aerodynamic force would be approximately 25% greater during flight, a scaling factor 
of 1.4 X was determined using this value combined with a measured maximum 
payload of 40 g for Robo Raven. Thus, this scaling factor made it possible to predict 
the corresponding payload capacity of Robo Raven from the wind tunnel data (Force 
Magnitude X Scaling Factor = Total Flight Weight).  The corresponding results, seen 
in Table 4, clearly explain why the original 22 module UAV did not fly given the 
predicted payload of -4 g, which was recovered to 21 g with the modified wing 
design. 
Table 4.4: Weight and payload characteristics for each UAV design 
 Robo 
Raven 
12 Module 
Robo Raven 
III 
22 Module 
Robo Raven 
III 
Modified 22 
Module Robo 
Raven III 
Weight of UAV (g) 290 317 331 346 
Force Magnitude (g) 234 235 232 260.1 
Total Flight Weight (g) 330 332 327 367 
Payload (g) 40 15 -4 21 
 
 Robo Raven 12 Module 
Robo Raven 
III 
22 Module Robo 
Raven III 
Modified 22 Robo 
Raven III 
 Residual 
Thrust 
(g) 
Lift 
(g) 
Residual 
Thrust 
(g) 
Lift 
(g) 
Residual 
Thrust (g) 
Lift 
(g) 
Residual 
Thrust (g) 
Lift 
(g) 
Trial 1 111 218 105 201 70.6 240 77.4 247 
Trial 2 104 220 98 219 75.2 242 101 229 
Trial 3 109 221 98 218 91.9 237 74.8 268 
         
Average 108 220 100 212 79.2 240 84.5 248 
Std. 
Dev. 
3.36 1.79 3.86 10.1 11.2 2.53 14.7 19.5 
  64 
 
 
4.3.2 3D Digital Image Correlation (DIC) Characterization of Wing 
Shape 
 
 Previously, Stanford et al. used 3D Digital Image Correlation (DIC) to study 
wing deformations in fixed membrane wings for MAVs to optimize their design for 
aerodynamic forces [40]. We utilized 3D DIC to study the effects of deformation on 
the different multifunctional wing designs by quantifying differences in shapes and 
strain and relating it to the aerodynamic loads generated during flapping. For our 3D 
DIC investigation, two Flea3 FL3-FW-03S1M cameras were used to acquire 
stereoscopic high speed images at 80 HZ while the wings were flapping at 4 Hz.  
Speckle patterns were applied to the surface of the wings, and the software package 
VIC-3D (Correlated Solutions, Inc) was used to obtain deformation measurements at 
20 different angles during a single flapping cycle. 
 
  65 
 
 
Figure 4.8: Comparison of out-of-plane displacement (W) for each wing at the 
horizontal position while flapping downward, which was found to be most 
representative of the relative deformations between the wing designs for the 20 
different wing positions that were measured during a single flapping cycle. (top left) 
regular, (top right) 6 module, (bottom left) 11 module, (bottom right) modified 11 
module. Dashed lines indicate approximate location of the modules 
 
 Representative DIC data associated with wing shape be seen in Figure 4.8, 
which was obtained by taking the out-of-plane displacement (W) in the z-direction 
normal to the wing while it is in a horizontal position while flapping downwards.  For 
these measurements, the x-axis and corresponding U displacements were chosen to 
run along the leading spar of the wings, while the y-axis and V displacements runs 
along the body of the UAV.  At the horizontal position, wings are generating the most 
aerodynamic lift and exhibit the greatest deformation.  It is clear that the regular wing 
has greater deformation towards the trailing edge of the wing than the 6 module wing 
or the 11 module wing, where the deformations are more indicative of bending on the 
  66 
 
leading edge.  These larger deformations are intuitive since the wings are not 
stiffened by the addition of solar cells.   
 
Figure 4.9: Comparison of time resolved residual thrust and shear strain: (top left) 
regular, (top right) 6 module, (bottom left) 11 module, (bottom right) modified 11 
module. 
 
 A comparison was also made between the four wings and the time-resolved 
resolved thrust and aerodynamic lift loads versus the average shear strain and biaxial 
strain respectively of the entire surface of the wing throughout the flapping cycle 
(Figures 4.9 and 4.10). The aerodynamic lift and thrust correlate strongly with the 
biaxial strain and shear deformation from the DIC results. Since we are interested in 
the change in biaxial strain, the mean strain was subtracted from the strain observed 
throughout the flapping cycle to compare the differences.  As mentioned earlier, it is 
clear that the integration of solar cells has an effect on the wing shape during the 
flapping cycle due to the increased stiffness of the solar cell material relative to 
  67 
 
Mylar. This in turn reduces the amount of thrust and lift as the solar cells are 
integrated into the regular wing design. However, by increasing the wing area in the 
modified 11 module wing design, residual thrust and lift could be recovered, although 
greater deformation was also observed that could influence performance.  The 17% 
increase in wing area provided an additional 7% of thrust force and 4% of lift force. 
The correlation between the lift and thrust forces to the biaxial and shear strains have 
been measured using the correlation coefficient. These values are shown in Table 4.5. 
 
Figure 4.10: Comparison of time resolved aerodynamic lift and biaxial strain relative 
to mean biaxial strain for the entire wing: (top left) regular, (top right) 6 module, 
(bottom left) 11 module, (bottom right) modified 11 module. 
 
 
 
 
 
 
  68 
 
Table 4.5: Correlation coefficients between thrust and lift forces and shear and 
biaxial strains respectively.   
 
 As the majority of the wing becomes covered in solar cells, the deformation of 
the wing decreases.  By observing the time resolved results from the 6 module and 11 
module wings, it is clear that the shear strain slightly decreases as solar cells are 
added.  Where the 6 module wing achieved a strain of 2% the 11 module wing 
remains under 2%.  By increasing the wing size and allowing for more deformation, a 
large increase in shear strain in the modified 11 module wing is observed.  These 
results are also mirrored in the cyclic results.  The shear strain for the 6 module wing 
and 11 module wings have a much lower value than the regular wings.  However, the 
modified 11 module wings have a much higher shear strain value.  This increase in 
deformation is the difference between the original 11 cell wing and the modified 11 
cell wing.  The increase in compliance is what allows the modified 22 module UAV 
to maintain flight.  The increase stiffness and weight of the solar cells is counteracted 
by the increase in overall wing deformation. 
4.4  Multifunctional Performance Modeling of Wings 
 
 A new model was developed to characterize the multifunctional performance 
of the wings based on the aerodynamics of flapping wing UAVs. Let Ft be the thrust 
generated by the flapping wings, V be the flight velocity due to Ft, and Fl be the 
 Robo 
Raven 
12 Module 
Robo Raven 
III 
22 Module 
Robo Raven 
III 
Modified 
22 Robo 
Raven III 
Thrust/Shear Strain  
Correlation 
Coefficient 
 
0.45 
 
0.39 
 
0.65 
 
0.66 
Lift/Biaxial Strain  
Correlation 
Coefficient 
 
0.56 
 
0.19 
 
0.45 
 
0.77 
  69 
 
aerodynamic lift at flight velocity V for the baseline UAV. Let Mm be the mass of the 
baseline UAV, and Mb be the mass of battery on the baseline UAV. To maximize the 
flight time, the largest possible battery permitted by the lift can be used, leading to the 
following condition: 
Fl = (Mm + Mb)g   (4.1)  
or       
Mb = Fl/g - Mm    (4.2) 
Let U be the energy capacity of battery on the baseline UAV. In general, U is 
proportional to the mass of the battery Mb. So,  
U = kbMb      (4.3) 
 
where, kb is battery coefficient. The flight time for this baseline configuration will 
therefore be: 
 
   T = U/P = kbMb/P = kb(Fl/g - Mm)/P   (4.4) 
 
Where P is the power consumed by the UAV during flight.  P is a variable that 
changes with each vehicle.  It is measured by observing the amount of time it takes to 
deplete a fully charged battery and knowing the battery’s energy capacity.   
For multifunctional wings, Ms is the mass of solar cells in the wings.  The 
solar cells have the following effects:  
 They are expected to alter the thrust due to stiffening of the wings.  The changed 
thrust leads to a different flight velocity, V’ , due to a relative change in drag 
force, k1 = V’/V= (F’t/Ft)0.5 
  70 
 
 They also change the aerodynamic lift coefficient by a ratio, k2, resulting in a total 
aerodynamic lift k1k2Fl.  
Equation (4.1) becomes, 
    k1k2Fl = (Mm + Ms + M’b)g   (4.5)  
 
Where M’b is the mass of battery on the UAV with solar cell integrated wings. So,  
 
    M’b = k1k2Fl /g - Mm - Ms   (4.6) 
 
Flight time for the multifunctional wings will therefore be as follows when directly 
providing power during flight:  
    T’ = kbM’b/ (P-ksMs)    (4.7) 
In this expression, the power produced by the solar cells is assumed proportional to 
the mass of the solar cell, and ks is the solar coefficient influenced by factors such as 
the conversion efficiency of the solar cell and the solar energy flux. Flight time for 
the multifunctional vehicle would be as follows if the solar cells do not provide power 
during the flight (i.e., baseline flight time): 
        T’ = kbM’b/ P    (4.8) 
 
Using Equation (4.6), we can determine the flight time when the solar cells provide 
power as follows: 
   T’ = kb (k1k2Fl /g - Mm - Ms)/(P-ksMs)   (4.9) 
 
  71 
 
Provided the current output of the solar cells does not exceed the recharge limit of the 
battery, the battery recharge time for an UAV with integrated solar cells can also be 
determined as follows: 
   Tr = fckb (k1k2Fl /g - Mm - Ms)/ksMs.   (4.10) 
 
Where fc is the fraction of the battery charge that was consumed before recharging.  
For most situations, flight time is considered a system design problem, 
resulting in a constraint on the minimum value of flight time, 
    Tflight ≥ Tmin    (4.11) 
 
If T ≥ Tmin, then there is no benefit in integrating solar cells into the wings such (i.e., 
T’= Tmin). Therefore, the objective in that case would be to select Ms such that Tr is 
minimized without T’ exceeding Tmin. Since there are complex interactions between 
the baseline wings and solar cells, solar cells should be placed such that k1 and k2 are 
maximized in order to maximize T’.  
Equation (4.7) enables the multifunctional criterion ks to be determined that 
allows for the mass of the solar cells to generate the same amount of current as 
consumed by the UAV (i.e., infinite flight time), and can be considered a critical 
multifunctional criteria, ks
*, under the following condition,   
               
                 
(4.12) 
  
  72 
 
Therefore, a comparison between the multifunctional criterion for the solar cells, ks
*, 
and the mass of the UAV, Mm, can also be made for design purposes based on: (1) 
area of solar cell coverage, and (2) power consumed by the UAV. 
4.5  Experimental Results for Multifunctional Performance 
 
 Inputs for the multifunctional performance model were obtained from 
experimental results. Since flight conditions can be different for each flight, these 
experiments were conducted while holding the vehicle in place and pointing the 
wings towards the sun.  These experiments were conducted on clear sunny days at 
noon with the planform area of the wings parallel to the sun.  The wings allowed to 
flap where Battery energy storage is typically stated as Ampere-hour (A-hr). Hence 
for the power calculations, energy storage is multiplied by the operating voltage of 
the battery and get power in terms of Watt-hour (W-hr).  The value kb is obtained by 
dividing battery capacity by battery mass, resulting in 106.54 W-hr/kg. To determine 
flight time, a fully charged battery was used to power the UAV until depleted by 
flapping, resulting in 4.50 min.  Using this in Equation (4.8), the average current draw 
from the batteries was found to be 4.9 A.  The UAV mass, battery mass, and solar cell 
module mass were determined using a DigiWeigh model DWP-1001 scale with 0.1 g 
resolution, resulting in 263.3 g for the UAV, 25.7 g for the battery, and 1.7 g per 
module. Therefore, the minimum total mass without solar cells is 289 g, and the Fl is 
2.83 N.  
 Due to the change in wing design for the modified 22 module UAV, the Mm 
increases to 278.3 g, resulting in a Fl of 2.98 N. The actual lift force generated by the 
UAV was previously measured to be 3.23 N.  To compare the aerodynamic lift 
  73 
 
changes caused by the integration of solar cells, the raw lift forces were compared for 
all wing designs.  Thus, the k2 was found to be 1.023 for the 12 solar cell UAV, 1.029 
for the original 22 module UAV, and 1.194 for the modified 22 module UAV.  
Similarly, k1 had to be calculated for each of the new wing designs since a change in 
thrust production was observed on the load cell.  Using the thrust forces observed on 
the load cell, k1 was 0.970 for the 12 solar cell UAV, 0.912 for the 22 module UAV, 
and 0.922 for the modified 22 module UAV. 
4.5.1 Direct Powering of Motors 
 
 To calculate T’, ks must be found first. Since ksMs gets subtracted from P, ksMs 
is equal to the power being supplied by the solar panels, which was measured to be 
4.10 W for the 12 module UAV.  Therefore, using the previously reported mass of the 
solar cells of 27 g, ks equals 0.152 W/g. This makes the new predicted maximum 
flight time 5.05 minutes. This represents an 11.5% increase in operation time using 
the solar cell wings. Thus, the overall effect of using the solar cell wings on flight 
time turns out to be positive despite the additional mass and rigidity that it adds to the 
wings.  
 The calculation is slightly different for the 22 module UAV.  Since the body 
of the UAV was modified to accommodate the new wing design, a different flight 
time can be expected since the weights of the UAVs are different.  To lift the 
difference in weight, the servos must pull more power from the battery, shortening 
the flight time.  The new flight time for this wing was 4.32 minutes, which is 
consistent with the increased weight and area requiring more energy to power the 
wings at the same frequency and amplitude.  The average current draw was calculated 
  74 
 
to be 5.14 A, which results in an average power consumption of 37 W.  Next, the new 
ks for the modified 22 module UAV must be calculated.  The solar cells add an 
additional 42 g to the UAV and were found to generate 7.41 W, making ks a value of 
0.176 W/g.  Using these numbers, we can predict the new maximum time of flight of 
the 22 module UAV using Equation (4.9).  The new time was calculated to be 5.23 
minutes.  This is 12.5 seconds more than the 12 module UAV and a 15.4% increase in 
operational time compared to the original Robo Raven design.  
 The prediction of the multifunctional performance model demonstrates the 
potential gain from the solar cells. However, it does not take into account the power 
generation variations introduced by flapping the UAV.  While flapping we expect a 
deviation from perfect conditions because the solar cells are constantly changing their 
position relative to the sun.  We experimentally measured the increase on the 
operation time by using both battery and solar cells to power motors. During actual 
flight tests with the solar cells electrically connected, large variations in flight times 
were observed due to variations in flight conditions.  Variations in flight conditions 
cannot be predicted or accounted for, so all flight endurance tests were done outdoors 
but on a test stand.  All outdoor tests were done on a clear sunny day at noon.  The 
wings were held at an angle where the most solar energy could be collected by the 
solar cells.  The actual operation time for the 12 solar cell UAV increased by 10.2% 
(flight time of 5.00 minutes) which was close to predicted.  The same test was done 
for the modified 22 module UAV.  The actual operational time for the 22 module 
UAV was 5.17 min, representing a 14.1% increase in operational time, which was 
  75 
 
also close to the prediction. Results for changes in power and flight time are 
summarized in Table 4.6. 
 It is important to note that while increasing the modules from 12 to 22 on the 
UAV only increased the operational time by 3.9%, the low increase was primarily 
due to the extra power required for the new wing design that reduced some of the 
solar cell benefit. This further reinforces the trade-off that is assessed when using 
more solar cells. Therefore, in addition to the time of flight, we also determined the 
critical ks
* according to Equation (4.12) which would require improving the solar cell 
production output instead of adding more solar cells or redesigning the wing.  Given 
Ms and P for the 12 module UAV, ks
* equals 1.34 W/g.  Given the value of 0.152 W/g 
for the flexible solar cells used in this investigation, we are at only 11.3% of the value 
needed for infinite flight. Thus, ~8.8X improvement is needed to reach infinite flight 
time.  Doing the same calculation for the Modified 22 module UAV, ks
* equals 0.787 
W/g.  With a ks
 value of 0.176 W/g, we are generating 22.4% of the power necessary 
for infinite flight.  Only ~4.5X improvement is needed for infinite flight.  Since the 
efficiency of the flexible solar cells we are using is only 5%, infinite flight time 
would only require increasing the efficiency to 22.5%, which would obviate the need 
for batteries and render them a secondary power source. 
  76 
 
Table 4.6: Comparison of predicted and measured flight time for regular and 
multifunctional wings (dnf denotes “did not fly”) 
 
 
 
 
 
 
 
4.5.2 Recharging of Batteries 
 
 To determine recharge performance to compare with the multifunctional 
performance model, the UAV was placed in sunlight, and measurements were taken 
to see how long it would take to completely recharge a depleted battery.  The 2 cell 
Lithium Polymer battery is completely recharged when it reads 8.4V.  The results for 
both the 12 solar cell and 22 module UAV are shown in Figure 4.11. It took 149 
minutes for the 12 solar cell UAV to completely recharge the battery and 90 minutes 
for the 22 module UAV.  These results are compared to the fastest theoretical results 
in Table 4.7 below. Differences can be attributed to the passive recharge circuit that 
minimized weight and power consumption. A more active circuit using maximum 
 Power 
Consumpti
on 
Solar 
Power 
Generation 
Predicted 
Increase in 
Flight Time 
Measured 
Increase in 
Flight Time 
 W W Time 
(sec) 
% Time 
(sec) 
% 
Regular UAV 35.8 n/a n/a n/a n/a n/a 
12 module 
UAV 
36.4 4.10 29.4 10.8 27.6 10.2 
22 module 
UAV 
36.8 7.41 dnf dnf dnf dnf 
22 module 
UAV 
(modified 
wings) 
37.0 7.41 41.9 15.4 38.1 14.1 
  77 
 
power point tracking could optimize recharge time, but at the expense of adding more 
weight and increasing the power requirements for the UAV. 
 
Figure 4.11: Recharging time profiles for the 12 and 22 module UAVs. 
 
 
 
 
Table 4.7: Comparison of theoretical charging to actual charging results. 
 Theoretical Fastest 
Recharging Time (Min) 
Actual Recharging Time 
(Min) 
12 module UAV 74.0 149 
22 module UAV 40.4 90 
 
4.5.3  Impact of Aerodynamic Forces of Power Output Solar Cells for 
Sensing 
 
The time-resolved power generated by the multifunctional wings was 
determined by measuring the voltage and current during flapping. In Figure 4.12a, 
results for the 22 module UAV, generating an average of 7.42 W, are compared with 
the 12 module UAV, generating 4.10 W.  The power output was expected to change 
  78 
 
sinusoidally while flapping, due to its position relative to the sun. However, the 
response is not sinusoidal, which indicates the deformation of the wings must have an 
effect on power output as well.  In Figures 4.12b and 4.12c, it was found that changes 
in the power output appear to directly correlate with the thrust generation of the wing. 
Since the thrust has been found to correlate with the amount of torsional deformation 
being produced, it is likely the cause of the changes is due to local rotations from the 
torsional deformations that can reorient cells more towards or away from the sun. 
Therefore, the solar cells wings are not only multifunctional in being able to harvest 
solar energy and serve as skin to generate aerodynamic force during flapping, but it 
can also be used to sense those forces due to the effects of the shear deformation.  The 
shear strains on the surface of the wings for both the 12 and 22 module UAVs were 
compared to the % change in power output to observe their correlation.  The two 
signals are compared in Figure 4.13.  The correlation coefficients for percent change 
in power output to thrust and shear strain were found to validate their relationship, 
and are shown in Table 4.8.   
  The implications of sensing thrust using solar cells has many applications 
going forward.  Gusts of winds can be detected during flights using the same structure 
that is used to help power the UAV.  Thus, this information can be used to change the 
flapping profile in reaction to the changes in aerodynamic loads.  These changes can 
be potentially automated to allow for correction while it is being piloted or flown 
autonomously, in which case the wings would be used as smart structures.  Because 
the solar cells are being used to achieve longer flights, being able to adjust to flight 
  79 
 
conditions using solar cell sensing can be a very powerful new tool for further 
increasing flight time.   
 
(a) 
 
(b) 
  80 
 
 
(c) 
Figure 4.12: (a) Electrical power generated by the 12 and 22 module UAVs while 
flapping, and thrust versus % change in power output for (b) 22 module and (c) 12 
module UAVs indicating that the solar cells can be used to sense aerodynamic forces 
due to the shear deformations. 
 
 
(a) 
  81 
 
 
(b) 
Figure 4.13: Shear strain versus % change in power output for (a) 22 module and (b) 
12 module UAVs 
 
Table 4.8: Correlation factors between thrust and % change in power output for each 
wing design 
 12 Module UAV 22 Module UAV 
Thrust/% Change 
in Power Output 
Correlation 
Coefficient 
 
0.87 
 
0.79 
Shear Strain/% 
Change in Power 
Output Correlation 
Coefficient 
 
0.53 
 
0.26 
 
4.6  Conclusions 
 
 This chapter investigates the mechanics that affect the potential benefits of 
introducing multifunctional structures to harvest solar energy on the wings of a 
flapping wing UAV.  We used Robo Raven as our base platform for this 
investigation.  Three different wing designs were initially observed: (1) the regular 
wings without solar cells, (2) wings with 6 solar cell modules that constitute a 12 
  82 
 
module UAV, and (3) wings with 11 solar cell modules that constitute a 22 module 
UAV.  Immediately, a deterioration in flight performance was observed when solar 
cells were added.  The 12 module UAV was still able to fly, but the wings of the 22 
module UAV were too stiff and heavy to generate enough aerodynamic lift, resulting 
in a calculated negative payload capacity.  Knowing that deformation plays a major 
role in force production, the wing design for the 11 module wings was altered, and a 
modified wing design was developed for the 22 module UAV to recover aerodynamic 
forces.   
 With these four wing designs, the force production of each wing design was 
compared to understand the mechanical effects the solar cells had on the aerodynamic 
performance of the wing.  Therefore, the lift and thrust forces generated by each wing 
design were quantified. Next, the deformation of each wing surface was quantified 
while flapping using 3D DIC to determine the specific effects of the mechanical 
properties of the solar cells as the wing design was varied. There was a clear 
correlation between the measured DIC deformations and the aerodynamic forces, in 
particular the correlation of the thrust force to the shear strain and biaxial strain to lift 
force which is critical to providing the forward velocity necessary to generate 
aerodynamic lift that enables the UAV to achieve flight.  It was also found that the 
shear deformation of the solar cells induced changes in power output that directly 
correlated to the thrust generation, indicating that the multifunctionality of the solar 
cell wings was not limited to just harvesting solar energy and serving as skin to 
generate aerodynamic forces, but that they could also be used to sense those forces. 
  83 
 
 Next, the electrical benefits of adding solar cells were determined. A charging 
circuit was developed so that the UAV can be charged by the solar cells.  The solar 
cells were also directly integrated to the electrical power of the UAV to extend the 
operating time of the UAV.  A performance model was developed to model the 
change in operating time due to the integration of solar cells. It was determined that it 
takes 149 minutes to completely recharge the battery with 12 solar cells and 90 
minutes to recharge with 22 modules.  Theoretically we could have a maximum 
13.9% increase in flight time with the 12 module UAV and a 21.2% increase for the 
22 module UAV.  Unfortunately the model does not take into account the flapping 
motion of the solar cells nor the heat of running these tests outdoors.  We observed a 
10.2% and 18.7% increase in operational time for the 12 module and 22 module 
UAVs respectively.  The current solar cells have an efficiency of 5%; however, with 
recent developments in flexible PV solar cell technology provide more efficient solar 
cells can be integrated to provide for a longer and maybe infinite flight time.  We 
found we would need a flexible PV solar cell that is 22.5% efficient to achieve 
infinite flight, and there are flexible PV solar cells that are more than 24% efficient 
commercially available. 
 In summary, the fundamental mechanics of adding solar cells to flapping wing 
UAVs has been established. The added stiffness of the solar cells prevents the wings 
from deforming resulting in a loss in aerodynamic force generation.  The shear and 
biaxial strains on the surface on the wing were directly correlated to the thrust and lift 
generated by the wings.  This new information is utilized in developing a 
multifunctional performance model to predict the effects on the flight time of the 
  84 
 
UAV. Although the stiffness and weight of the solar cells were quantified and 
determined to have a negative effect on the aerodynamic force production, this was 
offset by the power generation of the solar cells to result in significant gains in the 
flight time. From this model, further addition of solar cells to larger wings, the body, 
or the tail of a UAV can be expected to result in increased flight time.  However, the 
greatest benefit to flight time is expected to result from higher efficiency flexible 
solar cells, such as thin film GaAs multi-junction solar cells, where a 4.5X 
improvement in efficiency will theoretically result in infinite flight time and obviate 
the need for batteries other than as a secondary power source.  Finally, it was 
determined that multifunctional solar cell wings were capable of three functions: (1) 
lightweight flexible structure to generate aerodynamic forces, (2) energy harvesting to 
extend operational time, and (3) sensing of aerodynamic forces. 
 
  85 
 
 
Chapter 5:  Integrating Solar Cells into Flapping Wing Air  
  Vehicles for Enhanced Flight Endurance 
 
 This chapter investigates ways to redesign Robo Raven to significantly 
increase the wing area and incorporate more solar cells into the wings, tail and body. 
This was the work presented in the Journal of Mechanisms and Robotics [51].  
Increasing wing area allows for additional solar cells to be integrated, but there are 
tradeoffs due to the torque limitations of the servo motors used to actuate the wings as 
well changes in the lift and thrust forces that affect payload capacity.  These effects 
were modeled and systematically characterized as a function of the wing area to 
determine the impact on enhancing flight endurance. In addition, solar cells were 
integrated into the body and the tail.  The new design of Robo Raven generated a total 
of 64% more power using on-board solar cells, and increased flight time by 46% over 
the previous design.  They were also able to recharge batteries at a similar rate to 
commercial chargers. 
5.1 Introduction  
 
 Recently, Robo Raven was developed at the University of Maryland [38,42].  
It features independent wing control that allows the wing motions to be programmed 
to perform complex aerobatic maneuverers. A picture of Robo Raven can be seen in 
Figure 1. Its flight endurance was limited by the small battery pack used to power the 
vehicle, which was a 370 mAh 7.4V LiPo.  Robo Raven was able to continually flap 
for around 4.5 minutes before the battery was depleted and significant decrease in 
flapping frequency was observed. 
  86 
 
 
Figure 5.1: Robo Raven I [38] 
 
 Energy harvesting techniques can be implemented in FWAVs to meet the 
need for more available energy during flight and at rest, while increasing payload 
capacity by generating more lift can provide more energy storage.  As part of 
previous work, Robo Raven IIIv1 was developed, which took advantage of a large 
amount of surface area in the wings to integrate solar cells [27].  Through this 
integration, the wings became multifunctional structures.  The wings not only 
performed the function of providing the vehicle with the aerodynamic forces 
necessary for flight, but also provided electrical power for charging the battery. 
Initially, the Robo Raven III platform consisted of 12 solar modules integrated into 
the wing structure of the original Robo Raven design (Robo Raven IIIv1). These 
modules were integrated to the front most area of the wing, which minimized the 
changes in deformation of the flapping wings.  The solar cells produced 3.6 Watts and 
resulted in a 27 second increase in flight time with the ability to recharge the battery 
to 8.4V in 149 minutes.  To produce more power, more solar modules were integrated 
into the rest of original wings’ surface area.  This resulted in a different wing design 
  87 
 
that allowed the FWAV to maintain flight.  Ultimately, Robo Raven IIIv2 featured 22 
solar modules that increased the flight time by 38 seconds representing a 14% 
improvement in endurance [42]. 
 This version of Robo Raven III was also able to completely recharge the 
battery in 90 minutes. However, the integration of solar cells had an effect on the 
flight performance of the FWAV.  The additional mass and increase in stiffness to the 
wings affected the wing deformations, which were necessary to achieve the shape 
changes responsible for aerodynamic performance.  They also caused the motors to 
pull 2.2 W more than the original Robo Raven IIIv1 design, however an additional 
7.4 W was produced by the solar cells resulting in an additional 5.2 W of power 
generation.   
 With infinite flight time being the ultimate goal, two approaches have been 
considered: (1) higher efficiency solar cells, and (2) integrating more solar modules 
into the FWAV.  This paper focuses on the latter.  By increasing the size of the wings 
to provide room for more solar cells and integrating solar cells into the body and tail 
of the vehicle, the harvested energy could be increased significantly.  Integration of 
solar cells into the body should have minimal effects on the performance of the 
vehicle, whereas integrating solar cells into larger wings and the tail should have 
much more significant effect on vehicle’s aerodynamic performance.  This paper 
investigates ways to redesign Robo Raven to significantly increase the wing area for 
incorporating additional solar cells and adding solar cells into the tail and body. 
5.2  Increasing Solar Cell Coverage by Increasing Wing Area 
 
5.2.1 Designing and Building Larger Solar Wings 
 
  88 
 
 The carbon fiber spar arrangement utilized in this paper was based on a 
previously successful wing design [26].  This geometry allowed the wings to deform 
favorably under actuation.  This spar layout was shown to be scalable when it was 
used for a larger platform [38].  Using this spar arrangement provides a good starting 
point when making further changes to the wings.  The parametric design 
configuration is the same as shown in Chapter 3&4 and is shown again in Figure 5.2. 
It is repeated to show exactly how the designs differ.  The angle between the spars is 
kept the same as the previous Robo Raven designs. 
 
Figure 5.2: Parameters for wing design: S is the semi-span, C is the chord, and tn are 
the diameters of carbon fiber stiffening rods 
 
 The solar cell configuration for each wing design was also based on a 
previous successful design [47].  This design showed that keeping the rear of the 
wing free to deform allows the wing to maintain a shape under actuation that aid 
generation of aerodynamic forces.  Moreover, adding solar cells to the front most part 
of the wing has the least effect on the deformation of the wing than integrating them 
anywhere else.  When integrating solar cells into new wing designs, it is best to 
  89 
 
integrate them from the front spar of the wing and expanding towards the rear. This 
configuration allows for wing to deform under aerodynamic loading and enable 
generation of appropriate aerodynamic forces. 
 In a previously reported study, we have investigated how scaling the wing size 
affects the performance [42].  Wings of 133%, 166%, and 200% of the size of the 
original Robo Raven wings were evaluated.  An increase in performance was 
observed for the 133% and 166% wings but not for the 200% sized wings.  Based on 
these findings, the 133% size wings and 166% size wings were considered as 
candidates for solar cell integration.   
 By increasing the stiffness and mass through the number and location of the 
solar cells, the deformation is more constrained and less aerodynamic force can be 
generated. Thus, a limit to the endurance gains was found to exist for the Robo Raven 
IIIv2 wing design (shown in Figure 5.3) when integrating solar cells into the wings. 
Therefore, in order to harvest more energy, larger wings are needed to increase the 
number of solar cells. However, increasing the number of solar cells and the wing 
area will not only alter the aerodynamic forces generated by the larger wing size, but 
also affect the flapping characteristics (e.g., flapping frequency and amplitude) due to 
the torque limitations of the servo motors.  The maximum torque the servo can handle 
is 22 kg-cm.  The closer the servos operate to the maximum torque, the slower they 
become until they eventually stall. The wings designed are not large enough or heavy 
enough to stall the servos. However, they were large enough to reduce the motor 
speed. Thus, a slight adjustment in the programmed frequency was used to 
  90 
 
compensate for the slowing down of servos.  Table 5.1 summarizes the characteristics 
of the new larger wing designs used in this investigation. 
Table 5.1: Characteristics of the different wing designs 
Wing Design 
Semi-
Span 
(cm) 
Chord 
(cm) 
Wing 
Area 
(cm2) 
Mass  
(g) 
Solar 
Cells 
Robo Raven I 
Wing 
52.71 31.12 1017.3 15.3 No 
Robo Raven III 
v2 
52.71 34.57 1229.5 33.2 Yes 
A 62.23 35.56 1368.0 19.3 No 
AS 62.23 35.56 1368.0 45.2 Yes 
B 66.68 39.37 1627.6 25.4 No 
BS 66.68 39.37 1627.6 50.0 Yes 
 
  91 
 
 
Figure 5.3: Wings designed, built, and characterized to determine the effects of solar 
cell integration on flight endurance for Robo Raven 
 
 Wings AS and BS are geometrically just like wings A and B respectively, 
except they have solar cells integrated into them.  We use Powerfilm’s© MPT6-75 
solar cell modules. These are flexible 7.3 x 11.4 cm solar cell modules produce 50 
mA of current at 6 V for a power output of 0.3 W.  More efficient commercial solar 
cells are available, but they tend to be several orders of magnitude more expensive 
compared to the solar cells used in this investigation.  Cost considerations prohibited 
us from using expensive high efficiency solar cells in our experimental studies. Low-
cost solar cells can be considered as surrogate material for this study focused on the 
mechanical design.   
  92 
 
 Modifications had to be made to the solar cells to reduce the mass and the 
bending stiffness to be more compatible with the Mylar® membrane of the wings and 
tail. By removing the protective encapsulation on the solar cells, the bending stiffness 
of the solar cells was minimized, and the weight of the cells was reduced to 
approximately 1.8 g. The solar modules were connected to one another to make 
panels that fit on the wings.  A total of 30 modules were integrated into the wings.  A 
hole was made in the Mylar to fit the panels into the membrane of the wing.  Once in 
place they were attached to the Mylar membrane and the spars were put in place to 
complete the wing.  In Figure 5.3, the original wing designs for Robo Raven and 
Robo Raven IIIv2 are compared to the newly developed wings A, AS, B, and BS. 
5.2.2 Modeling Performance Changes Based on Design Changes to the 
Wing 
 
 One important parameter for flapping wing flight is the Strouhal number [49]. 
It is a dimensionless number that is used to describe oscillating flow mechanisms.  
The formula for the Strouhal number (St) is shown in equation (5.1). 
 
          (5.1) 
 
Where f is the flapping frequency, A is the flapping amplitude in meters, and vf is the 
forward velocity in m/s.  It was found that a Strouhal number between 0.2 and 0.4 
provided the most efficient flight for animals in nature [48].  The first design of Robo 
Raven was slightly under this metric.  By increasing the Strouhal number, more 
efficient flight can be achieved.  However, the servo motors used to actuate the wings 
constrained the available power to drive the wings. Increasing the size and mass of 
wings will cause the flapping amplitude to decrease for a given motor, as well as the 
  93 
 
move the motor operation point to the low efficiency regime. Smaller amplitudes for 
flapping are detrimental to thrust production, since thrust relies on the volume of air 
that can be moved by the wings in the direction opposite to the forward velocity. In 
turn the velocity of any point on the wing moving through the air is a major factor in 
thrust production.  The equation for the magnitude of the velocity profile, v, of the 
rotating wing is: 
 
       (5.2) 
 
Where f is the flapping frequency, r is the distance any given point on the wing from 
the axis of rotation, and 𝜃 is the angle of the wing during the flapping cycle.  This 
equation was adapted from the angular velocity of the wing and described in terms 
that can be used in predictive models.  The velocity is limited by the motor torque 
motor.  The power effects for the drag of the wing through air and against gravity 
were adapted from Wu et al. [28] and can be calculated as follows: 
 
        (5.3) 
 
Where CD is the drag coefficient of the wing, ρ is air density, A is the wing area, m is 
the mass of the wing, cg is the center of gravity for the wing measured from the motor 
that actuates the wing, and ω is the angular velocity of the wing.  However, power 
effects are expected to change under different operating conditions and for different 
wing designs. As described in previous findings [38], they can be easily transformed 
when the axis of rotation is reoriented to the angle of attack during flight. Therefore, 
the power effects can be accounted for when the velocity, mass of the wing, center of 
gravity, and flapping angle change for each wing design.   
  94 
 
 Our previous studies on compliant wing structures have shown that there is a 
direct correlation between spar deformations and thrust [25]. In these studies, 
analytical models of the wing shape and aerodynamic forces generated (FL and FT) 
due to drag during flapping have been employed and adapted from previous work, 
and then were modified to include aerodynamic lift as follows [26]: 
 
       (5.4) 
 
        (5.5) 
 
where CL is the coefficient of lift, S is the planform area of the wing, and vf is the 
forward velocity of the platform. These equations describe the forces generated when 
the axis of rotation is in the direction of thrust due to the drag of the wing as it rotates, 
along with the lift generated as it is dragged forward through the air during flight. 
This model produced good correlation with observed trends in lift and thrust, which 
conformed to assumed characteristics of the wings generating these loads. Therefore, 
these models can be used to predict the changes in lift and thrust production for each 
wing design due to the integration of solar cells.   
 In addition, empirical terms have been employed for the thrust to account for 
the compliance of the wing while the wing captures air during its deformed state and 
pushes air towards the rear of the vehicle at the apex and nadir of the flapping cycle, 
known as the “blowback effect”, as follows [38]: 
 
       (5.6) 
 
                          (5.7) 
 
  95 
 
where  is the change in total drag force that is decomposed into corresponding 
contributions to lift and thrust as given in Equations (5.4) and (5.5) based on the angle 
of attack, k is an empirical constant of proportionality previously determined to be 
0.2, f is the flapping frequency, θ is the angle of the wing during the flapping cycle, 
Df is the level of the drag force above the point at which the deformation transitions 
from global to local during blowback, c is the global stiffness of the wing, δ is the 
displacement of the mid-chord of the semi-span of the wing if it were infinitely rigid, 
and vo is the corresponding rigid wing velocity. It is important to note that the 
contribution from   to thrust and lift only occurs when the wing is near apex or 
nadir [25] and corrects for the duty cycle on force generation for the platform.  
5.2.3 Wing Testing 
 
 Each wing was tested using a force measurement system similar to the one 
introduced in the previous chapter to determine the differences in force production. 
Using the same 6 degree-of-freedom (DOF) load transducer, the relevant lift and 
thrust forces associated with flight were measured simultaneously.  Changes to the 
previous test stand were made to measure the lift and residual thrust generated by the 
vehicle.  This test stand is set higher to compensate for the larger wings tested.  The 
Delrin platform was also switched for a wooden structure to further eliminate 
vibrational effects.  The test stand allows the vehicle to be pitched to a 20° incline to 
replicate flight conditions.  The entire test stand is placed at the end of a wind tunnel 
to measure aerodynamic lift and residual thrust under actual flight conditions.  The 
wind tunnel was set up to replicate the flight conditions of Robo Raven and its 
  96 
 
inflight Reynolds number of 120,000.  The test stand and its components can be seen 
holding the FWAV in Figure 5.4. 
 
Figure 5.4: Test stand with residual thrust and lift directions identified. The ATI 
Mini40 6 DOF load cell is capable of measuring up to 40 N of force with a resolution 
of 0.01 N in the thrust direction and 120 N of force with a resolution of 0.02 N in the 
lift direction 
 
 In addition to measuring the aerodynamic forces generated by the wing during 
flapping, the test stand is also capable of measuring the position of the wings using an 
optical encoder to characterize the flapping range.  All of the signals are 
simultaneously collected using a National Instruments NI DAQ-9188 data acquisition 
module.  The raw data is sent to a LabVIEW VI that converts the raw signal to the 
actual loads being measured, and then the final loads are saved to an Excel template. 
 To effectively characterize the change in performance between the wing 
designs, tests were conducted over a range of frequencies with no tail attached.  
Larger wings were expected to perform better while flapping slower than the original 
wings, which flapped at 4 Hz.  Therefore, each wing design was tested at 3, 3.25, 3.5, 
3.75, and 4 Hz.  The frequencies were measured in real time using the encoder on the 
  97 
 
test stand and the LabVIEW VI software.  The frequency was adjusted manually 
using a remote control until the desired frequency was achieved.  A minimum of 3 Hz 
was required to maintain flight. A significant drop in flapping amplitude, and 
subsequent lift and thrust force generation, was observed beyond 4 Hz.  To ensure the 
wings were operating at their peak performance, the FWAV was powered by a 
constant voltage power supply.  The load cell measurements were then compared 
with model predications. 
 The load cell results from testing Robo Raven in a wind tunnel at speeds of 6 
m/s were found to be representative of force production during actual flight after 
using a correction factor related to the difference in the actual flight conditions [47].  
Integrating solar cells into Wing AS added 51.8 g, compared to its counterpart 
without solar cells (Wing A). The previously reported payload for Wing A was 46.8 g 
[42]. Therefore, the only way for Wing AS to maintain flight is if it could produce 
higher lift forces than Wing A.  Wing BS weighed 49.2 g more than Wing B. Wing B 
had a previously reported payload of 76.6 g [42]. Therefore, Wing BS However could 
underperform Wing B by 27.4 g, and still maintain flight.  The results from load cell 
testing are shown below and are compared to the predicted results from the wing 
models in equations (5.4) and (5.5). 
  98 
 
 
(a) 
 
(b) 
  99 
 
 
(c) 
Figure 5.5: Load cell results and subsequent changes in payload capacities compared 
to predicted results from models for different wing designs: Top) Residual Thrust, 
Middle) Aerodynamic Lift, Bottom) Difference in Payload. Mass of each single wing: 
Wing A = 19.3g, Wing AS = 45.2g, Wing B = 25.4g, Wing BS = 50.0g 
 
 As the flapping frequency is increased from 3 to 4 Hz, the thrust decreased for 
each wing design. Wings with solar cells produced the least amount of thrust, 
however Wing BS produced the least amount of thrust where its non-solar powered 
counterpart, Wing B, produced the most amount of thrust.  For each frequency that 
was tested, Wing A produced the least amount of lift, Wing AS the second least, 
Wing B the second most, and Wing BS produced the most amount of lift.  The largest 
lift forces were observed at 3.5 Hz. 
  As previously noted, the weight of solar cells integrated into Wing A was 
beyond the inherent payload capacity of the wing design, so the increase in lift was 
  100 
 
barely sufficient to achieve flight, as seen in Figure  5.5.  On the other hand, the 
weight of solar cells added to Wing B was less than the inherent payload capacity of 
the wing design, and with the measured increase in lift and thrust, a total of 48.1 g of 
additional payload capacity was achieved by integrating solar cells into the Wing B 
design at 3.5 Hz.  Further evidence of these effects was also observed during actual 
flight tests, where Wing AS was not able to climb and maintain flight while Wing BS 
was able to.   By observing the Strouhal number for each wing design under each 
flapping frequency, the best overall flapping performance can be predicted that is 
consistent with the lift and thrust measurements.   
Table 5.2: Strouhal number for each wing design under each flapping condition 
 
Strouhal Number 
3 Hz 3.25 Hz 3.5 Hz 3.75 Hz 4 Hz 
Wing A 0.192 0.198 0.224 0.220 0.199 
Wing AS 0.175 0.179 0.182 0.180 0.178 
Wing B 0.194 0.199 0.203 0.204 0.208 
Wing BS 0.175 0.183 0.192 0.190 0.184 
  
 The models were also able to predict how changes in the wing design affected 
the vehicle performance.  By using the actual recorded amplitude in the predictive 
models, the limitations imposed on wing performance by the available torque of the 
motors were simulated.  The trends were all consistent with the actual trends observed 
with the measured data. The changes in thrust were larger than expected.  The 
predictions for the wings with solar cells were lower than the actual measured values, 
where the predictions for the wings without solar cells were over the measured 
values. The lift model was slightly different.  Where the predictive model suggested 
that the largest lift force would be observed by Wing BS at 4 Hz, it was actually 
measured with Wing BS at 3.5 Hz.  This showed that we are reaching a torque 
  101 
 
limitation in the servo motors that is accounted for by changing the flapping range 
used to calculate lift and thrust in the models.  The models were able to predict that 
integrating solar cells would ultimately decrease the thrust produced and increase the 
lift.  When solar cells are integrated into the wings, they stiffen the wings causing the 
wing deformations to decrease, subsequently reducing the amount of air the wings 
could capture and move towards the rear of the vehicle to produce thrust.  On the 
other hand the increase in inertial forces due to solar cell integration can affect the 
aerodynamic forces generated during “transitions” in the flapping direction where a 
“whipping” effect that increases the contribution to both lift and thrust forces. The 
additional weight of the wings will also increase the moment acting on the motors, 
resulting in a great power draw from the servo motors while rotating the wings at a 
given frequency. 
5.3  Increasing Solar Cell Coverage by Utilizing the Tail 
 
5.3.1  Modeling Performance Based on Design Changes to the Tail  
 
 Integration of solar cell into the tail allows for several different tail designs to 
be developed and characterized.  The solar modules chosen to be integrated to the 
body and tail are the same as those used for the wings.  Based on the 676 cm2 surface 
area of the original tail, it was possible to integrate a maximum of 4 solar cell 
modules.  To provide uniform charging of the lithium polymer battery from each 
module, an even number of solar modules had to be introduced to the system.  On the 
body there was enough space for 2-3 more solar cell modules.  With this constraint, a 
maximum of 6 additional solar cell modules could be integrated into the system, since 
  102 
 
2 modules are required in series to generate the 8.4 V necessary to charge the battery 
or operate the servo motors. 
 In addition to the wings, the tail also plays a crucial role in flight performance.  
Actuated by a small servo, it allows the vehicle to turn left and right.  It also plays a 
role in flight stability, allowing the vehicle to maintain its pitch through the moment 
generated by the drag force acting on the tail.  Therefore, integration of solar cells 
into the tail is expected to affect aerodynamic performance of the FWAV.  The tail’s 
effective drag changes with geometry.  This alters the vehicle’s ability to convert 
thrust force into aerodynamic lift.  Two main design changes were identified that 
would allow for the tail to keep its current functionality while making it possible to 
integrate solar cells.  First, the surface area of the tail can be increased. This change 
will cause more drag, and is expected to reduce thrust while increasing lift.  However, 
it also allows for more solar cells to be integrated to the tail.  The second aspect is the 
geometry of the tail, which can have its spread angle altered.  This allows for 
different orientations of solar module integration to be tested, and for determining the 
effects of changing the drag moment by changing the width of the tail while 
maintaining a constant surface area. 
 In order to determine the effects of the tail design on the aerodynamic forces 
and moments generated during flight, it was necessary to develop an appropriate 
analysis.  Using previously developed models for downwash effects on lift and thrust 
[48], predictive equations can be developed to describe tail effects on aerodynamic 
forces.  Assuming that the vehicle has a constant forward velocity (i.e., steady state 
flight condition) and no pitch, the changes caused by the tail on the thrust force 
  103 
 
generated by the flapping wings, FtT, can be predicted from the drag and lift acting on 
the tail surface area by partitioning it. One portion of the wing, βAT, will be in the 
blowback area of the wake of the flapping wings where the velocity increases to vf+af 
proportion to the flapping frequency f. The other portion, (1-β)AT will be near the 
centerline of the body not exposed to the blowback, and experiences the normal 
forward velocity of vf. The flapping frequency factor, a, can be determined by varying 
the flapping frequency. The portion of the tail exposed to the wake, β, is determined 
by flapping with and without air flow.  Therefore, we can calculate the total change in 
the thrust force due to the tail as follows: 
     (5.8) 
where Cd is the drag coefficient determined from conventional drag equations  for a 
plate with aspect ratio, AR, as  follows:  
    (5.9) 
where 𝜃T is the pitch of the tail on the body of the FWAV, 𝜃0 is the angle of attack at 
which the drag goes to 0, and e is the efficiency factor of the shape. Therefore, a taller 
tail will have a lower value of β, and a wider tail will have value closer to 1.  
Similarly, the contribution to the lift force from the tail, FlT, can also be derived 
similar to the drag force produced by the tail and the increase from the blowback 
effects from the wings: 
     (5.10) 
where the coefficient of lift, Cl, is given by: 
     (5.11)  
  104 
 
5.3.2 Designing and Fabricating Solar Tails 
 
 To validate the effects of these design changes and the model, new tails with 
solar modules were designed and fabricated.  Tail geometries were limited by the size 
of the solar modules integrated into tail.  The new tails were designed to be as small 
as possible while still being able to integrate 3 or 4 solar modules.  The 120° design 
was the largest angle that enables 3 modules to fit inside and the 90° design was the 
smallest angle that enabled 4 modules to fit inside the tail while still having a similar 
surface area to the 120° 4 module design.  These four designs allowed us to 
characterize how tail geometry affects FWAV performance.  The design 
specifications for each new tail design can be seen in Table 5.3 and a schematic of 
each geometry is shown in Figure 5.6.   These tails were tested using the same force 
measurement system as the wings.  However, to understand the drag effects of each 
tail design, one extra set of data with the vehicle parallel to the ground was collected. 
Table 5.3: Specifications for solar cell tail designs 
Tail Design 
Specifications 
Surface 
Area 
(cm^2) 
Angle 
(degrees) 
Mass 
(grams) 
Number 
of Solar 
Cells 
Original (0) 354 120 6.2 0 
1 678 90 16.0 4 
2 546 90 13.1 3 
3 683 120 17.0 4 
4 546 120 14.0 3 
  105 
 
 
Figure 5.6: Solar cell tail designs that were tested: a) Tail 1 b) Tail 2 c) Tail 3 d) Tail 
4 (dimensions in cm) 
 
 Each of these designs is expected to have different effects on vehicle 
performance. However the number of solar modules introduced to the FWAV system 
was constant because the voltage requirements limited the addition of solar cells to an 
even number. Thus, the new tail designs contribute up to 1.2 W of additional power, 
increasing the maximum power that can be generated to 10.8 W, or a 12.5% increase. 
5.3.3  Tail Testing 
 
 The results from wind tunnel testing gave new insights to the benefits of each 
tail design.  The lift and thrust loads produced by each tail design under the 4 
conditions were found with a wind velocity of 6 m/s, which is consistent with the 
velocity of the FWAV during flight.  By comparing the drag at 0°, thrust at 20°, lift at 
20° we can characterize the role of tail in the vehicle flight. 
 The static response of each tail is consistent with the physics of drag and the 
model.  A larger tail is expected to create a larger drag force in the wind tunnel.  That 
is evident when comparing the tails that simply differ in the amount of solar modules 
integrated.  If the geometry of the tail is the same, the tail with more solar modules 
  106 
 
and a larger surface area will create more drag.  This can be observed in the drag 
results for each tail design presented in Figure 5.7.  The 120° angle spread tails 
generate more drag than their 90° spread counterparts.  Tail 1 and Tail 4 have the 
same amount of solar modules but the 120° spread tail generates more drag.  The 
same observation can be made when comparing Tails 2 and 3, but with 4 solar 
modules instead of 3.  The 90° tails have longer spars that give the vehicle a larger 
height, thus the aerodynamic forces on the taller tails create larger drag moments 
causing the spars to bend into a more aerodynamic shape.  This aerodynamic shape 
and bending means that less surface area is orthogonal to the incoming airstream, 
creating less drag for the vehicle, which is why the drag for Tail 1, the tallest tail 
design tested, is so low compared to the predicted model. 
 
Figure 5.7: Comparison of the drag from each tail design while wings are stationary 
and vehicle at a 0° angle of attack 
 
 The static response of the vehicle while completely horizontal was observed 
for each tail design and measured.  However, in actual flight the FWAV has a pitch of 
  107 
 
20°, and additional drag forces are generated by the tail because of the blowback from 
the flapping wing that results in thrust.  The best way to understand these effects is to 
run the same test with the FWAV at a pitch of 20° with the wings flapping at 3.5 Hz.  
Since the wings are flapping, a net positive residual thrust force is expected to 
overcome the drag.  This residual thrust exists because the air speed of the wind 
tunnel is not high enough to equate the average thrust force generated.  The effects of 
the wings for these tests can be seen in Figure 5.8.  It is clear that 90° tails have less 
of a drag effect on the FWAV while flapping.  This is due to the tails being narrower 
and capturing less of the air generated by the flapping wings.  All of the 120° spread 
tails experience a larger drag force, thus generating a smaller thrust force.  When 
comparing just the 90° tails, the effects from surface area are intuitive.  The larger 90° 
tail (Tail 1) causes a larger drag force, thus a smaller residual thrust is observed than 
the smaller 90° tail (Tail 2).  The opposite seems to be true for the 120° tails.  The 
smaller tail (Tail 4) with a smaller surface area and fewer solar modules has the 
higher drag force than its 4 module counterpart (Tail 3).  This can even be seen with 
the original tails with no solar modules, where it generates the largest drag force.  The 
larger 120° tail is more susceptible to the air blown back by the wings flapping 
allowing the tail to bend slightly more than the smaller wing.  This more streamline 
shape causes less drag at a 20° pitch.  The model was able to take the geometry of the 
tail and predict what would happen to the thrust force for each of these changes (see 
Figure 5.8). 
  108 
 
 
Figure 5.8: Comparison of the residual thrust from each tail design while wings are 
flapping and vehicle pitched at 20° 
 
 The thrust forces with the FWAV at a 20° pitch are clearly shown.  Given the 
position of the vehicle and that incoming air is present the thrust result make sense.  
The tail plays a major role in lift generation.  The tail of the FWAV is positioned at a 
45° incline relative to the body.  During flight, incoming air generates a moment with 
the tail that causes the FWAV to have a pitch of 20°.  Tails of different shapes, sizes, 
and design are expected to have an effect on lift generation.  The lift results, seen in 
Figure 5.9, help to further explain the effects of these different designs.  When 
comparing the tails against each other they are the complete opposite to the thrust 
results.  Excluding the original tail, the tail that provided the smallest thrust actually 
provided the most lift. Tail 3 and 4 were both 120° spread tails and generated the most 
lift.  Since these tails are wider, they are able to capture more of the air generated by 
the wings flapping.  The same effect that caused more drag and reduced thrust 
assisted in generating more lift. This is what the model predicted and this is exactly 
what was seen by the measured results. 
  109 
 
 
Figure 5.9: Comparison of the lift from each Tail Design while wings are flapping 
and vehicle is pitched at 20 degrees. Mass of each tail: Tail 0 = 6.2g, Tail 1 = 16.0g, 
Tail 2 = 13.1g, Tail 3 = 17.0g, Tail 4 = 14.0g 
 
 Wind tunnel testing provides a comparison amongst the five tail designs. This 
new information allows improvements in actual flight performance and payload can 
be estimated.  Using the known payload with the original tail, the payload can be 
estimated for the new tail designs. These predictions can be seen in Table 5.4.   
Table 5.4: Payload calculation from total force generation 
Tail 
Design 
FWAV Mass 
(g) Including 
Battery 
Total Force 
Generated (g) 
Estimated 
Total Flight 
Weight (g) 
Estimated 
Payload (g) 
 
0 346 131 367 21 
1 364 137 373 9 
2 362 153 389 27 
3 365 153 389 25 
4 363 161 397 34 
 
 Each of the new tail designs should be capable of generating enough force for 
flight.  In fact, an increase in payload of up to 57% can be expected from all of the 
new tail designs except for 1.  Tail design 4 was found to produce the greatest energy 
increase with 120% more than the original tail design 0 that did not have solar cells. 
  110 
 
These estimations were confirmed during actual flight tests, where it was observed 
that the vehicle was able to fly with each of the tail designs.  From load cell tests, 
integrating Tail 4 for our design would produce the best results and was the tail 
chosen for integration. This version of Robo Raven III with the same 22 solar module 
wings as Robo Raven IIIv2 and solar cells integrated into the tail and body of the 
vehicle was named Robo Raven IIIv3. 
5.4 Increasing Solar Cell Coverage by Utilizing Body 
 
 Integration of solar cells into both the wings and tail are expected to have 
significant effects on the flight performance of the FWAV. However, integrating 
solar cells into the body should have a minimal effect on vehicle performance. The 
integration of solar cells into the body should simply add mass to the system, which 
takes away from the payload capacity, and does not affect the force generation or 
aerodynamics of the wing and tail structures.  To integrate solar modules into the 
body, it was necessary to design two mounts that allow solar modules to attach to the 
carbon skeleton of the FWAV’s body, but allow as much area as possible to orient 
towards the sun. 
 Tail mount that houses the tail servo was re-designed to provide a surface for 
the solar cells to be attached to the top of the vehicle.  Another mount was designed 
as a simple clip with the same surface to hold the other end of the solar cell panel. 
Depending on the configuration of solar modules used for the tail (3 or 4), 2 or 3 
modules were used to create the solar panel to be integrated with the two body 
mounts. The new mounts for the solar cells added 4.4 grams to the vehicle.  The two 
module configuration adds a total of 7.8 grams to the FWAV and the three module 
  111 
 
configuration adds 9.5 grams to the system (1 solar cell module = 1.7g). The 
integration of the 6 new solar modules to the body and tail of Robo Raven IIIv3 can 
be seen in Figure 5.10.   
 
Figure 5.10: Two different design configuration for integrating 6 additional solar cell 
modules: (Top) 2 modules in body, 4 modules in tail; (Bottom) 3 modules in body, 3 
modules in tail 
 
5.5 Results and Discussion 
 
5.5.1 Flight Tests 
 
 The fabricated wings and tails were all flight tested to validate the results from 
load cell testing.  Each wing design that did not have solar cells was tested first to 
determine their fundamental flight capability.  The best metrics to observe the best 
flight were the vehicle’s climb rate and total flight times.  A design with a fast climb 
rate shows good aerodynamic force production and affords a higher payload capacity.   
  112 
 
Consistent with load cell results, it was determined that Wing B flew the best.  It 
clearly had a faster climb rate than Wing A.  However, the first new phenomenon 
observed was that the larger wings seemed to operate better while flapping slightly 
slower than the smaller wings where they could achieve the desired flapping range 
with the torque-limited servo motors. 
  Next, the wing designs with solar cell were tested. It was determined that 
Wing AS was not able to maintain flight due to an inability to achieve sufficient lift to 
maintain an adequate climb rate.  This flight behavior was consistent with the results 
from load cell testing, because Wing AS produced only slightly more lift than Wing 
A, which was inadequate to carry the additional 51.8 g from the solar cells.  However, 
Wing BS were able to produce enough extra lift and thrust to maintain flight.  The 
total payload for these wings must have only been slightly greater than the 51.8 g of 
the solar cells, because there was a notable decrease in overall vehicle performance.   
 Flight testing the new tail designs with the Robo Raven III v3 wings was 
consistent with the results observed from load cell testing.  First, the FWAV was able 
to fly with each new tail design. However, the worst flight was with Tail 1.  It simply 
did not climb like the others.  It had a notably slower climb rate.  Due to changes in 
wind and condition variability, it was not possible to observe many more differences 
in performance between the tail designs.  However, from our load cell results Tail 4 
was chosen to be the best design.  Using Wing BS and Tail 4 designs, Robo Raven 
IIIv4 was built and flown, and is shown in Figure 5.11. 
  113 
 
 
Figure 5.11: Robo Raven IIIv4 
 
 This final Robo Raven IIIv4 design was thoroughly flight tested to determine 
its flight characteristics.  With the improvements that went into this design, a total of 
36 solar modules were successfully integrated into the wings, tail, and body of this 
FWAV.  This final design was able to maintain flight.  The FWAV weighed 361.4 
grams without the battery.  With the battery, the vehicle weighed 388 grams.  The 
vehicle was able to carry an extra 6 grams before it would no longer climb.  The flight 
characteristics of Robo Raven III v4 are shown in Table 5.5 below and compared to 
the original Robo Raven design.   
 
 
 
 
 
 
  114 
 
Table 5.5: Flight Characteristics of Robo Raven and Robo Raven IIIv4 
Flight 
Characteristics 
Robo 
Raven [26] 
Robo Raven 
IIIv4 
Vehicle Mass 263.3 g 361.4 g 
Forward Velocity 7 m/s 5.6 m/s 
Climb Rate 0.53 m/s 0.23 m/s 
Turning Radius 6.1 m 5.5 m 
Payload (without 
batteries) 
54 g 32.6 g 
 
 As seen in Table 5.6, the FWAV with 36 solar modules produced 10.8 W. 
However, since the wings were larger, they required more current from the battery.  
Where the previous version took 4.5 minutes to deplete the battery, the larger wings 
took 3 minutes and 45 seconds due to the aforementioned increase in torque required 
to drive the heavier wings.  The 36 module FWAV flapped for 4 minutes 35 seconds. 
This 50 second increase demonstrated a 22.2% increase in total flight time, which is a 
46% increase over the previous design.  It is important to note that these 
measurements were conducted at the highest flapping frequency of 4 Hz (due to 
limitations of the remote control used outdoors), and that lowering the flapping 
frequency decreases power consumption and increases flight time while also 
generating optimal aerodynamic force at a frequency of 3.5 Hz as previously 
reported. 
 
 
 
 
  115 
 
Table 5.6: Flight Time Results for Different Numbers of Solar Cell Modules 
Robo Raven 
Version 
Number of 
Solar Cell  
Modules 
Maximum 
Power 
Generated (W) 
Increase in 
Flight 
Time(sec) 
% increase 
Robo Raven I 0 0 0 0 
Robo Raven III v2 22 6.6 38 15.2 
Robo Raven III v4 36 10.8 50 22.2 
 
5.5.2  Battery Recharge Tests 
 
 While wind tunnel testing provided new insight into the change in flight 
performance for different design changes, the electrical benefits of adding solar cells 
to the FWAV needed to be characterized as well.  The solar modules were expected 
to extend the flight time of the FWAV and decrease the recharge time for the battery.  
Outdoor tests were conducted to determine the effects that the 36 modules will have 
on harvesting solar energy.   
 To measure flight endurance, the FWAV was flapped outdoors until the 
battery was completely depleted.  The FWAV was held in place with the normal 
surface of the wings pointed directly at the sun before the test began.  This ensured 
that the photovoltaic cells were exposed to the maximum amount of sunlight for 
maximum power production.  Upon plugging in the battery, the FWAV started 
flapping and a timer was started.  The timer was stopped as soon as the FWAV 
stopped flapping due to the lack of power.  These times were recorded and compared.  
To measure the recharging time of the FWAV a recharging circuit was built to safely 
recharge the battery to full charge.  A zener diode with a breakdown voltage of 4.3 V 
was used as a voltage regulator. This value is necessary as the charging voltage of the 
Lithium Polymer battery is limited to 4.2 V per cell. The resistor in the circuit is 
  116 
 
placed to dissipate power so a 1.3 W zener diode can be used instead of a higher 
wattage.  This circuit is designed under the assumption that the solar panel acts 
primarily as a current source at constant voltage. A schematic of the recharging 
circuit used to test the recharging capabilities of the 36 module design is shown in 
Figure 5.12. 
 
Figure 5.12: Recharging circuit that regulates voltage going to the battery for safe 
recharging where 4.2V are supplied to the battery at the blue node and 8.4V are 
supplied at the red node 
 
 The 2 cell Lithium Polymer battery uses a three node plug for safe charging.  
However, the consequence of charging a single battery using this recharge circuit is to 
reduce the effective charging current from 600 mA to only 300 mA. Furthermore, the 
maximum power output is also not maintained since the voltage is decreased to a 
limit of 4.2 V. This obviates the advantage of using a Maximum Power Point 
Tracking circuit. However, commercially available MPPTs have a negative effect on 
FWAV performance by increasing weight beyond the payload capability. In this 
circuit, the battery is charged only when the D2 diode has a minimum of 4.2 V across 
it and D1 has a minimum of 8.4 V. Current flows from high voltage to low voltage, so 
  117 
 
the solar cells must always be higher than the voltage of the battery.  The diodes are 
capable of reaching the minimum voltage even on a party cloudy day. It is expected 
that on an overcast day the diodes would not give enough solar power to reach the 
required voltage. Testing was conducted with a measured solar flux of 102 mW/cm2 
(i.e., full sun) in order to prove the solar cells were capable of charging the battery. 
 
Figure 5.13. Comparison of recharging results 
 
 The total recharging time of the 22 module FWAV was previously found to be 
73 minutes.  The new 36 module FWAV was able to recharge a depleted battery in 44 
minutes.  More solar cells allow for more current to be transferred to the batteries.  
With a power increase of 4.2 W, or 64% more power, we were able to recharge faster 
than a typical LiPo battery charger (Figure 5.13).   
5.6  Conclusions 
 
 This chapter investigates the tradeoff in flight performance when changing the 
design of wings and tail in FWAVs to integrate solar cells for harvesting energy and 
enhancing flight endurance. New wings were designed to successfully fly with the 
  118 
 
amount of solar cells integrated.  To utilize all the surface area available, solar cells 
were integrated to the body and tail of the vehicle.  Even though the effects of 
integrating the solar cells to the body were negligible, the effects of integrating solar 
cells to the tail and wings were significant.  
 Predictive models were implemented and shown to predict how changes to the 
wings and tail can affect performance.  These models can be used for other FWAVs 
that are forward flying and rely on a deforming membrane to generate thrust. Using 
larger wings allows greater forces to be generated.  However, integrating solar cells 
increases the inertial effects of the wings and stiffens the wings. Therefore, 
appropriate design of wings to achieve the necessary deformations during flapping 
was determined to be key to producing adequate thrust and lift for flight performance 
when solar cells are integrated into the wings.  Thus, a properly designed wing that is 
larger and heavier due to solar cell integration was shown to achieve better flight 
performance than one that is smaller and lighter, while also providing the wings with 
the ability to harvest solar energy. 
 This study also investigated improving performance through altering the 
FWAV tail design and integrating solar cells into them.  It was determined that the 
spread angle between the spars had the greatest effect on aerodynamic forces, where a 
tail spread of 120° will produce more lift and less thrust than a tail with a 90° spread 
of the same surface area due to these tails capturing more of the air pushed back by 
the flapping wings.  Based on how a FWAV is performing, the tail can be altered to 
accommodate the necessary tradeoff in lift and thrust that is characteristic of the tail 
design as follows:  (a) a wider tail can be used to generate more lift from faster 
  119 
 
moving air associated with the blowback effect from the wings, while (b) a narrower 
tail can be used to generate more thrust by avoiding this effect. 
 The resulting Robo Raven IIIv4 design produced 10.8W of the 37W needed 
for continuous flight during sunlight.  The flexible Amorphous Silicon cells used are 
only 5-6% efficient. Flexible  Gallium Arsenide cells that are 28-30% efficient would 
produce a total of 54W using the same vehicle design while only adding 5g to the 
total vehicle weight.  Realizing this design with GaAs cells would allow us to achieve 
flight using only solar cells during sunlight.   
 Lessons learned from this study will be applicable to the broad FWAV 
community. Wing deformation significantly affects the production of aerodynamic 
forces.  Integration of solar cells should be done to minimize the negative impact on 
the wing deformation during the flapping cycle.  Our studies show that it is better to 
integrate solar cells towards the front spar of the wing. Bigger wings help in 
harvesting more energy, but they also need more power to flap at higher frequencies, 
as evidenced by a significant drop in flapping amplitude and subsequent lift and 
thrust generation.  The approach described in this chapter can be used to find the right 
size of solar integrated wings for a given actuator. 
  120 
 
 
Chapter 6:   Designing a Flapping Wing Air Vehicle Capable of 
    Continuous Solar Flight by Integrating High  
    Efficiency Photovoltaic Solar Cells 
 
 This chapter is completed work that has yet to be submitted.  Up to this point, 
flexible solar cells have been integrated into the wings, body, and tail of our flapping 
wing air vehicle.  The solar cells have allowed the vehicle to stay in flight for longer 
periods of time and allows for the battery to recharge without using a battery charger.  
However, the technology used in the previous versions of Robo Raven III use 
amorphous silicon solar cells that have an efficiency of 6%.  Using flexible high 
efficiency solar cells, creating a FWAV that stays in flight using only solar power is 
possible.  This paper investigates more efficient photovoltaic technologies and 
establishes GaAs photovoltaic solar cells as the best option going forward. A model 
that predicts the mechanical performance due to a change in stiffness and material 
used in the wings was developed.  A new wing design using a material similar to the 
GaAs commercially available as developed, tested, and flown.  The new wing design 
was lighter and more deformable than the previous version allowing for better 
performance.  By actually integrating the GaAs solar cells in place of the mock 
material, the first FWAV capable of flying continuously in sunlight can be developed. 
6.1 Introduction 
 
 A major drawback of flapping winged flight is the continuous need of power 
to keep the vehicle flapping. Typically payload is extremely limited since flapping 
winged flight is challenging to achieve and the vehicle must be as light as possible to 
maintain flight. With a limited payload, the battery that is used to power the vehicle is 
  121 
 
typically small.  This small battery can only store so much energy which usually 
results in short flight times.  Shorter flights restrict the UAVs overall mission 
capabilities.  If the vehicle could harvest energy while in flight, the UAV would have 
longer flight times.  Harvesting enough energy would allow the UAV to stay in flight 
indefinitely.     
 The previous versions of Robo Raven III demonstrate the development of 
solar cell integration into our Robo Raven vehicle. Robo Raven III v1 was an 
investigative study to see if solar cell integration was possible.  By achieving flight 
and increasing flight time, we knew that solar cell integration was a viable way to 
increase flight time.  To investigate the limitations of solar cell integration, the solar 
cell wings were redesigned for Robo Raven III v2.  This design consisted of 11 solar 
modules integrated to each wing.  This was the maximum amount of solar cells that 
could be integrated into the original Robo Raven wings and still maintain flight.  
Robo Raven III v3 consisted of the same wings developed for Robo Raven III v2 
however, with solar cells integrated into the body and tail of the vehicle.  This 
allowed 6 more solar cells to be integrated into the vehicle.  This also demonstrated 
just how important the tail of the FWAV is for generating lift. The different tail 
designs tested gave new insight to how the tail translates the thrust generated by the 
wings into aerodynamic lift. 
  122 
 
 
Figure 6.1: Previous Robo Raven & Robo Raven III design: Top left: Robo Raven I, 
Top right: Robo Raven III v1, Bottom left: Robo Raven III v2, Bottom right: Robo 
Raven III v3 
 
 With Robo Raven III v3, the total available surface area of the original Robo 
Raven design was completely saturated with A-Si solar cells.  However, with 28 A-Si 
solar modules, the FWAV was harvesting a maximum of 8.4W while it was 
consuming 37W. For completely solar powered flight the FWAV would need to 
harvest 37 or more Watts.  Using the same solar cell technology, the best option was 
to increase the available surface area for solar cell integration.  By designing larger 
wings for the FWAV we can increase the surface area while also generating more 
thrust and lift.  A previous study [42] observed the effects of using larger wings for 
Robo Raven.  It was found that larger wings can actually generate larger forces while 
flapping at a slower rate than the current design. However, going too large can be 
detrimental since the flapping is limited to the torque the motors can output.  Another 
study was done where solar cells were integrated to larger wings for a Robo Raven III 
design that would carry the maximum amount of solar cells [51].  The largest wings 
  123 
 
that allowed the most amount of solar cells to be integrated were found.  These wings 
had a wingspan that was 66% larger than the original Robo Raven wings.  This 
allowed a total of 30 A-Si solar modules to be integrated into the wings.  With the 
additional solar cells on the body and tail of the FWAV, a total of 36 solar modules 
made up Robo Raven III v4.  Producing a total of 10.8W, we are producing 29% of 
the energy we need for completely solar flight.  At this point, the limits of A-Si 
technology has been reached.  New more efficient materials must be investigated to 
help reach the goal of completely solar flight.  However, integrating different 
materials, instead of the A-Si solar cells that are currently integrated, is expected to 
alter the way the wings deform, thus changing the performance of the FWAV.  A new 
investigation must be done where different solar cell technologies are tested and their 
effects on wing performance are estimated, measured, understood, and modeled. 
6.2 Solar Cell Technologies 
 
6.2.1  Amorphous Silicon Solar Cells 
 
 For photovoltaic solar cells to be a viable method for energy harvesting, thin 
flexible solar cells must be used for the wings to deform correctly and generate the 
lift and thrust necessary for flight.  Currently Amorphous Silicon (A-Si) solar cells 
are being integrated into the wings.  The specific solar modules chosen for integration 
were provided by Powerfilm Inc. They provide several off-the-shelf models to choose 
from.  To steadily increase the number of modules integrated into the body while still 
providing an appropriate voltage, model MPT6-75 was chosen.  This module had an 
operating voltage of 6V and an operating current of 50mA.  Each module can cover 
an area of 72.5mm by 111mm.  The A-Si modules have a protective encapsulation to 
  124 
 
protect the surface of the solar cells.  With the encapsulation each module weighed 
2.0 grams and had a thickness of 0.18mm.  However, the encapsulation made the cells 
too stiff when compared to the Mylar membrane of the wings.  By removing the 
encapsulation, the stiffness of the solar cells was reduced as well as shedding a 
considerable amount of the module’s total weight.  Each module now weighed 1.6g 
while the thickness was reduced to 0.14mm.  This was a 20% drop in mass and 22.2% 
drop in thickness.  These solar cells have proven to be a material that can be 
integrated into the wings of a FWAV.  However, producing a FWAV that would fly 
continuously in sunlight would be impossible with the low amount of electrical 
energy produced with these cells.  These A-Si cells were found to be 6% efficient 
under the best conditions.  While the MPT6-75 modules were the perfect candidate 
for exploratory research, their lackluster electrical performance influenced further 
investigations into more efficient photovoltaic technology.   
 
Figure 6.2: Powerfilm MPT6-75 A-Si Solar Module 
 
  125 
 
6.2.2  Semi-flexible Polycrystalline Solar Cells 
 
 The next photovoltaic solar cells considered for integration into FWAVs were 
polycrystalline solar cells.  These semi flexible solar cells were 17.6% efficient.  The 
solar cells are silicon based and the silicon material is melted then poured into a mold.  
As the silicon cools, it crystalizes in an imperfect manner.  The grains of the crystals 
can actually be observed on the surface of the solar cells.  On the back side of the 
cells is a thin piece of aluminum that serves to electrically connect each of the cells 
together. These cells are not as efficient as monocrystalline solar cells, but are more 
flexible.  These solar cells came in various sizes but the cells we observed were 8cm 
x 15.5cm and weighed 6g.  If the A-Si solar cells were simply swapped out for the 
polycrystalline solar cells, the FWAV would be harvesting 31.7 Watts of the 37 
needed for continuous solar flight.  However, these solar cell feel much stiffer than 
the A-Si cells and proved very difficult to integrate.  When these cells were integrated 
into the wings, they shattered during the wing fabrication process. By being very 
careful and gentle during the wing fabrication process, we were able to complete a set 
of wings with these solar cells. However, when it came to flapping these wings, the 
cells shattered while flapping. 
  126 
 
 
Figure 6.3: Polycrystalline Solar Cell 
 
6.2.3  Gallium Arsenide Solar Cells 
 
 The final solar cell technology that was considered for integration were 
gallium arsenide (GaAs) photovoltaic cells.  The manufacturer use a metalorganic 
chemical vapor deposition process to grow a thin layer of GaAs,  then remove this 
thin layer via an epitaxial lift-off (ELO) process which leaves a thin, flexible, and 
lightweight solar cell. These thin film flexible solar cells can be up to 30.8% efficient 
if they are dual junction, however the single junction cell observed was 24% efficient. 
The flexibility of these cells are more comparable to the flexibility of the A-Si cells.  
Comparing the efficiencies of these two cells, completely solar flight could not only 
be achieved by these solar cells, but with less surface area than the total surface area 
of solar cells in the current Robo Raven III v4 design.   A sample GaAs module 
produced can be seen below.  However, custom GaAs cells can be ordered for 
specific voltages and shapes. 
  127 
 
 
Figure 6.4: Sample GaAs photovoltaic solar module 
 Through previous investigations, the effects of integrating the A-Si cells is 
well known; however, integrating other solar cell materials will be slightly different 
due to their difference in mass and stiffness.  The physical properties of these 
materials must be compared so that the change in performance can be predicted.  A 
stiffness coefficient must be found for each of the materials to understand how much 
more or less the wings will deform while flapping.  The density of the cells must also 
be found to determine the change in weight of the vehicle and wings.  To determine 
the stiffness coefficient, the elastic modulus of each material was found using a micro 
tensile testing machine.  A sample was cut from all of the solar cell materials as well 
as the Mylar material that makes up the membrane of the Robo Raven’s wings.  The 
surface of the sample was painted in a speckle pattern.  This machine applied a 
continuous tensile load on the sample while a 2D Digital Image Correlation system 
measure the strain on the speckled surface of the sample.  This enables us to measure 
  128 
 
the force and strain on sample.  Using the cross-sectional area of each sample, the 
Elastic modulus was calculated by observing the elastic region of the results.   
 The thickness of all of these materials will play a major role in stiffness.  
However, the Mylar was the only material that was isotropic.  The other materials 
consisted of layered materials and the weakest layer of that material would cause the 
failure.  Below are the physical characteristics of the different materials that were 
investigated. 
Table 6.1: Comparison of photovoltaic materials 
Material Thicknes
s 
(mm) 
Elastic 
Modulus 
(GPa) 
Stiffness 
Coefficien
t 
 
Volumetric 
Density 
(kg/m3) 
Density 
by 
Area 
(kg/m2) 
Mylar 0.025 3.2 0.08 1390 0.0353 
A-Si Cell 0.14 1.7 0.24 1420 0.199 
Polycrystalline 
Cell 
0.23 7.1 1.63 2110 0.486 
GaAs Cell 0.14 1.5 0.21 1580 0.221 
 
 Analyzing these results, it becomes quite clear which high efficiency solar 
technology would be the best fit for the next Robo Raven III design.  The 
polycrystalline cells were 6.8 times stiffer than the A-Si cells and 2.4 times heavier 
per unit surface area.  The integration of the polycrystalline solar cells would clearly 
have a negative effect on performance when compared to the performance of Robo 
Raven III v4.  The large increase in mass will make it more difficult for the motors to 
actuate the wings.  The large increase in stiffness will not allow the wings to deform 
when the deformation is necessary to produce the thrust and lift forces to maintain 
flight.  The only way to integrate polycrystalline cells into the FWAV is to limit the 
integration to just the body and tail of the vehicle. 
 
  129 
 
6.3  A-Si and Polycrystalline Powered Vehicle 
 
 To determine the benefits that using polycrystalline photovoltaic cells can 
potentially bring, a FWAV was designed that incorporated them into the body and tail 
of Robo Raven III.  As previously stated, Polycrystalline cells were initially 
integrated into the wing; however, through the fabrication process these cells began 
cracking and falling apart.  Even when the wings were fabricated without breaking 
the solar cells, upon initial testing, the forced deformation of the wing induced by 
flapping caused the cells to crack and eventually completely fracture.  Integrating 
these cells into the wings was simply not an option.   
 To stay consistent with the aerodynamic performance of the previous Robo 
Raven III v4 vehicle [51], the same wing geometry was chosen for the new 
polycrystalline tail.  To sufficiently cover a similar area, two different sized 
polycrystalline modules were integrated into the tail. The first were 2 large 15x8cm 
modules that can each generate 1.9W. The second were 8 small 5.8x1.9cm modules 
that each generated 0.14W.  One additional large polycrystalline module was 
integrated to the body of the FWAV.  These polycrystalline solar cells alone generate 
6.82W of power.  Including the 30 A-Si modules already integrated into the wings, 
this platform generates 15.82W.  The new tail design can be seen in Figure 6.5. 
  130 
 
 
Figure 6.5: Polycrystalline tail and body integrated to Robo Raven III v4 
 
 Outdoor testing was performed to understand how the increase in power 
translates to an increase in flight time.  First, the FWAV was held in place and 
oriented so that the solar cells produced the most power.  The vehicle was first 
flapped with just the battery integrated to determine the flight time without the solar 
cells.  Then, the vehicle was flapped with the solar cells to determine the benefit the 
solar cells provide.  Without the solar cells, the vehicle flapped for 3 minutes and 35 
seconds.  With the solar cells the vehicle flapped for 4 minutes 45 seconds.  This was 
10 seconds more than the 36 A-Si FWAV.  The previous design demonstrated a 22% 
increase in power for the vehicle where the new design that includes the 
polycrystalline cells had a 33% increase in power.  This vehicle generated the most 
power out of any Robo Raven III design manufactured and can be seen in Figure 6.6. 
  131 
 
 
Figure 6.6: Robo Raven with polycrystalline solar cells in the tail and body, and 
Powerfilm flexible solar cells in the wings. 
 
 The benefits in power generation are clearly there; however, the 
polycrystalline cells added a substantial amount of weight compared to the previous 
design with the A-Si cells.  Robo Raven III v4 only had a payload of 32.6g [51].  
When the battery is considered, this is only an extra 6g of payload capacity.  The new 
tail and body weighed 11.1g more than the previous design.  This increase in mass 
proved to be detrimental during actual flight tests. The FWAV could not maintain 
flight simply due to the increase in mass.  From this it was concluded that the benefits 
provided by the polycrystalline cells did not outweigh their negative effects on flight. 
 The best path forward is to use the GaAs solar cells.  The GaAs cells are less 
stiff than the A-Si cells.  This means that the integration of these solar cells is 
expected to be similar to the integration of the solar cells used for Robo Raven III v4.  
The GaAs cells are slightly denser than the A-Si cells but they are so much more 
  132 
 
efficient that they will not take the same surface area to achieve completely solar 
flight.   
6.4  Modeling of the Effects of Wing Characteristics on Flapping Power, 
Thrust, and  Lift 
 
 Integrating different materials into the wing structure changes the 
aerodynamic forces the wing can produce.  It changes how the wings deform and 
therefore how much air the wings can interact with during the flapping cycle.  For this 
vehicle, the design of the flapping wing plays a major role in force production.  The 
planform area of the wing, weight of the wing, and stiffness of the wing are expected 
to alter the performance of the FWAV.  The weight of the wing will alter the flapping 
amplitude and frequency of the wing.  Applying a larger load makes it more difficult 
for the motors to actuate the wings.  This result in a slower flapping rate with a 
smaller amplitude.  The velocity profile of the flapping wing, vo, is given by the 
following equation [51]. 
      (6.1) 
Where f is the flapping frequency, r is the distance any given point on the wing from 
the axis of rotation, and 𝜃 is the angle of the wing during the flapping cycle. This 
equation was adapted from the angular velocity of the wing and described in terms 
that can be used in predictive models.  The velocity is limited by the motor torque 
motor. 
 The power effects for the drag of the wing flapping through air and against 
gravity were adapted from Wu et al. [45] and can be calculated as follows: 
       (6.2) 
  133 
 
Where CD is the drag coefficient of the wing, ρ is air density, A is the wing area, m is 
the mass of the wing, cg is the center of gravity for the wing measured from the motor 
that actuates the wing, and w is the angular velocity of the wing.  However, power 
effects are expected to change under different operating conditions and for different 
wing designs. As described in previous findings [38], they can be easily transformed 
when the axis of rotation is reoriented to the angle of attack during flight. Therefore, 
the power effects can be accounted for when the velocity, mass of the wing, center of 
gravity, and flapping angle change for each wing design. 
 Our previous studies on compliant wing structures have shown that there is a 
direct correlation between spar deformations and thrust [25]. In these studies, 
analytical models of the wing shape and aerodynamic forces generated (FL and FT) 
due to drag during flapping have been employed and adapted from previous work, 
and then were modified to include aerodynamic lift as follows [38]: 
      (6.3) 
       (6.4) 
where CL is the coefficient of lift, S is the planform area of the wing, 𝜃 is the angle of 
attack, and vf is the forward velocity of the platform. These equations describe the 
forces generated when the axis of rotation is in the direction of thrust due to the drag 
of the wing as it rotates, along with the lift generated as it is dragged forward through 
the air during flight. This model produced good correlation with observed trends in 
lift and thrust, which conformed to assumed characteristics of the wings generating 
these loads. Therefore, these models can be used to predict the changes in lift and 
thrust production for each wing design due to the integration of solar cells. 
  134 
 
 In addition, empirical terms have been employed for the thrust to account for 
the compliance of the wing while the wing captures air during its deformed state and 
pushes air towards the rear of the vehicle at the apex and nadir of the flapping cycle, 
known as the “blowback effect”, as follows [38]: 
       (6.5) 
                    (6.6) 
where  is the change in total drag force that is decomposed into corresponding 
contributions to lift and thrust as given in equations (4) and (5) based on the angle of 
attack, k is an empirical constant of proportionality previously determined to be 0.2, f 
is the flapping frequency, θ is the angle of the wing during the flapping cycle, Df is 
the level of the drag force above the point at which the deformation transitions from 
global to local during blowback, c is the global stiffness of the wing, δ is the 
displacement of the mid-chord of the semi-span of the wing if it were infinitely rigid, 
and vo is the corresponding rigid wing velocity. It is important to note that the 
contribution from   to thrust and lift only occurs when the wing is near apex or 
nadir [25]. 
The global stiffness of the wing will change with different solar cell materials 
and allows for the change in lift and thrust to be predicted.  Based off Equation (6.6), 
increasing the global stiffness will decrease the flapping velocity of the wing. This 
decrease in velocity decreases the aerodynamic forces given in Equations (6.3) and 
(6.4).  The global stiffness can be found using the following equation that follows the 
linear rule of mixtures: 
     (6.7) 
  135 
 
where km and ks are the stiffness coefficients of the Mylar membrane and solar cell 
material respectfully and Am and As are the percent area of the wing taken up by the 
mylar and solar cell material.  Using this global wing stiffness in the velocity 
equation (6.6) allows the lift and thrust forces to be predicted for different wing 
designs. 
6.5  Wing Design for Integrated GaAs Solar Cells 
 
 Integrating the GaAs solar cells into wings follows the same process as the A-
Si cells except slightly easier.  The manufacturer can develop custom photovoltaic 
solar cells to meet the needs of Robo Raven. They can provide a cell module that 
operates at 7.68V while providing a 233mA current.  Each module measures 
141.8mm x 50mm and provides 1.712 W.  This means we will only need 22 GaAs 
solar modules to reach the 37W needed to completely power Robo Raven IIIv5.  That 
is a total area of 0.156m2 that needs to be taken up by solar cells on the vehicle.  Robo 
Raven III v4 has 0.290m2 taken up by solar cells with 0.241m2 on the wings.  With 6 
of the new GaAs cells being integrated into the tail and body, only 16 modules need 
to be integrated into the wings.  That is a total area of 0.113m2 of the wings that needs 
to be taken up by GaAs solar cells.  That is 53% less surface area than is taken up by 
the A-Si cells in Robo Raven III v4’s wing design.  Since less surface area of the 
wing is taken up by the solar cells, this new wing design will go through larger 
deformations while flapping than the Robo Raven III v4 design. 
 Not only do these modules take up less surface area, preparing them for wing 
integration is much easier.  The A-Si modules came with an encapsulation that 
needed to be removed to reduce the module’s stiffness and mass.  The manufacturer 
  136 
 
of the GaAs cells is able to send the custom modules without any encapsulation.  This 
eliminates a very time consuming process in wing fabrication. 
 To build these new wings, the GaAs cells must be available.  Due to the high 
demand for this newer more efficient technology, using a mock material instead of 
the actual GaAs cells is preferable.  The best option was to use a material that had a 
similar or higher stiffness coefficient with the same density.  If flight can be achieved 
with a stiffer material, then flight can be achieved with the GaAs cells; however, the 
same density is required to replicate the load on the motors and total payload of the 
vehicle.  The material used as the mock material was 110 copper.  A 0.025mm thick 
roll was cut into the same shape as the GaAs cells.  The copper shim was about twice 
the stiffness of the GaAs cells, but had the exact same density as the solar cells.  If the 
new wings can achieve flight with these mock copper cells, then they would be able 
to achieve flight with the actual GaAs cells. 
 To actually make the new wings, the layout of the solar cells must be 
determined.  Keeping the solar cell at the front most section of the wing allows for the 
best performance [47].  Using the same size and geometry of the Robo Raven III v4 
wings, 8 mock GaAs cells can be easily integrated into the front most portion of the 
wings.  This design calls for the 8 modules to be laid in two horizontal rows.  When 
complete this wing looks very similar to a larger version of Robo Raven III v1.  
Interestingly though this design will harvest enough energy to completely power the 
FWAV. This wing weighed 37.2g which was 21.5g less than Robo Raven III v4’s 
wing.  A picture of the completed wing can be seen in Figure 6.7. 
  137 
 
 
Figure 6.7: Left: Mock GaAs wings Right: Completed Robo Raven III v5 design 
 
 With the additional 6 cells on the tail and body of the FWAV, this vehicle 
mimics a solar cell powered FWAV.  If this vehicle were to have the actual high 
efficiency GaAs solar cells, it would have the capability to stay in flight as long as 
there is sunlight.  However, this vehicle must be able to maintain flight.  Comparing 
this vehicle’s performance to the performance of Robo Raven III v4 would show a 
clear understanding of the effects of the vehicle’s performance and the effects of 
changing the design and material used for solar cell integration.   
6.6  Performance of Mock GaAs Wings 
 
 To quantify the difference in performance that these changes to the wing 
cause, the previous wing performance was compared to the performance of the new 
design.  The forces generated by each wing design were measured and compared.  
The same test stand with a 6 degree of freedom from Chapter 5 was used to measure 
the forces produced by the wings.  This 6 DOF load cell is capable of measuring up to 
40 N of force with a resolution of 0.01 N in the thrust direction and 120 N of force 
with a resolution of 0.02 N in the lift direction.  It allows us to simultaneously record 
the lift and thrust forces being produced. The raw signal is sent through a Labview 
  138 
 
program that converts the raw voltage to forces and the data is platted in an Excel file.  
The raw time resolved results can be seen below. 
 
Figure 6.8:  Time resolved results of Robo Raven III v4 and v5 
 
 To compare the difference in performance the average forces produced by 
each wing design were observed.  These lift and thrust forces are necessary to 
maintain flight and can be predicted using the models mentioned before.  The results 
demonstrate that the new Robo Raven III v5 wings produce more thrust and less lift 
than Robo Raven III v4 as was predicted by the models.  This was expected because 
the ability for the wing to deform increased causing more thrust to be generated while 
flapping.  However, the greater mass of the Robo Raven III v4 wing allows more lift 
to be generated during the downwards portion of the flapping cycle.  
  139 
 
 
Figure 6.9: Average lift and thrust generated and compared to the predicted results 
 
 From these measured results, a prediction can be made about the actual flight 
results.  Like airplane or a fixed wing aircraft, this type of FWAV relies heavily on 
the production of thrust to generate lift.  However, this vehicle has the advantage of 
also generating lift by flapping.  Ideally to measure the actual lift force each vehicle is 
capable of producing, the average thrust force should reach 0g.  The extra thrust 
produced by Robo Raven III v5 would translate to more lift.  How much is unclear 
for this new design.  The best way to discover the lift is to run payload tests with the 
new design.  Actual flight tests are the final real test to directly compare the two 
designs.  The results from actual flight tests can be seen below. 
 
Table 6.2:  Flight Characteristics of Robo Raven III v4 and v5 
 Total 
Vehicle 
Mass (g) 
Forward 
velocity 
(m/s) 
Climb Rate 
(m/s) 
Payload 
(g) 
Robo Raven III v4 381 5.6 0.23 32 
Robo Raven III v5 340 7 0.55 42 
  140 
 
 Robo Raven III v5 had an additional 10 grams of payload.  Even though the 
wing from Robo Raven III v4 generated more force, it had to overcome a larger 
vehicle mass.  The payload measured did not include the battery necessary for flight. 
If the payload included the 27 gram battery, the payload actually goes down to 5 
grams for Robo Raven III v4.  Since this FWAV developed had a material that 
mimicked the actual high efficiency solar cells, a battery had to be used to power this 
vehicle.  However, with the actual high efficiency GaAs cells, only a very small 
battery or large capacitor must be used to safely help power the vehicle.  The payload 
will still be close to 40 grams with the GaAs cells.  Not only will this FWAV fly 
longer than its predecessor, it will actually have better performance as well. 
6.7  Expected Electrical Impact of New Design 
 
 The combination of wing design and solar cell technology in Robo Raven III 
v5 is expected to outperform any previous version and produce enough energy 
through solar power to completely power the vehicle.  The percentage of the wing 
covered by solar cell material and efficiency of the solar cell material play a major 
role in the amount of power that can be produced.  The following plot demonstrates 
the potential of the wing design used. 
  141 
 
 
Figure 6.10: Potential Power produced by Robo Raven III v5 wing design 
 
 Ignoring the fact that the maximum theoretical photovoltaic efficiency is 86% 
and a wing that is completely covered by would never fly, this wing design can 
produce 327.5 Watts of electrical power. However, we are limited to the efficiency of 
thin film solar cell technology.  Figure 6.11 demonstrates the potential of the A-Si 
cell and the GaAs cell. 
 
Figure 6.11: Power produced by available solar cell technology for Robo Raven III 
v5 wing design 
 
Robo Raven III v4 
Robo Raven III v5 
  142 
 
 Needing only 37 Watts to power the entire vehicle, the new wing design with 
GaAs cells requires less than half of the total wing area to produce that power.  Using 
the A-Si cell, this goal would never have been reached. The Robo Raven III v4 wing 
used more than 50% of it wing area and was still capable of flight. This means that 
The GaAs version will be able to fly.  It will also be able to rely solely on solar power 
to fly. 
6.8  Conclusions 
 
 Current battery technology has limitations when applied to FWAVs.  The 
necessity for a small lightweight battery limits how much energy can be stored to be 
used during flights.  This means that typical flights and missions for FWAVs can only 
span a short distance for a limited amount of time.  Previous energy harvesting efforts 
have proven that solar cells can be integrated to the wings, tail and body of FWAVs 
and more specifically Robo Raven.  Previous versions of Robo Raven III used 
amorphous silicon cells.  These cells allowed for investigations of how wing 
deformation is affected by solar cell integration; however, lacked the efficiency to 
provide prolonged or completely solar flight.  
 This investigation sought to understand the integration of high efficiency solar 
cells into an already established viable design.  Different solar cell technologies were 
observed and gallium arsenide cells were determined to be the best material to be 
integrated to the wings.  With a similar stiffness and density to the A-Si cells, not 
much difference was expected in terms of performance.  Using this stiffness, a model 
that was previously developed was adapted to include a change in performance 
  143 
 
caused by changes in stiffness to the wing.  This model allowed the performance of 
the new wing design to be predicted. 
 Without being able to acquire the GaAs solar cells, copper shim material was 
used as a substitute material.  This copper shim was the same density, but twice as 
stiff as the solar cells.  This meant that if the new wing design was capable of flight 
with this material, it would be capable of flight with the GaAs cells.  The increase in 
efficiency from the A-Si cell to the GaAs cells is dramatic enough that less of the 
win’s surface area is used to harvest all of the energy needed to power the FWAV 
while in flight.  The new design called for 16 GaAs modules to be integrated to the 
front most portion of the wing, 4 modules in the tail, and an additional 2 modules in 
the body.  The final completed design was named Robo Raven III v5. 
 Load cell testing was done to compare the performance of Robo Raven III v5 
to the previous design (Robo Raven III v4).  It was found that the new design 
outperformed the previous version with respect to thrust but underperformed with 
respect to lift.  The increase in deformation is what caused this increase in thrust.  The 
additional mass of the previous version allowed it to achieve a larger force in lift.  
Even though the previous version generated more lift it also had a larger mass to 
overcome to achieve flight.  The new Robo Raven III v5 had a smaller overall mass 
and had a larger payload capacity than Robo Raven III v4. Taking into account the 
weight of the battery, Robo Raven III v4 had a total payload capacity of 5g.  Since 
Robo Raven III v5 does not need a battery, it has a total payload of 40g.  This means 
that Robo Raven III v5 mechanically outperforms the previous version.  By actually 
  144 
 
integrating the GaAs cells into the wings of this final design.  The first FWAV with 
the capability to fly indefinitely with sunlight will be developed.   
  145 
 
 
Chapter 7:  Conclusions 
 
 There were several contributions to the field of FWAVs that came from this 
work.  Although this work can be seen as pioneering in terms of solar cell integration 
to different structures of FWAVs, there were many small contributions along the way 
that gave new insight to the field in general.  These contributions include new testing 
methods, new fabrication methods, and intellectual contributions. 
7.1  FWAV Fabrication 
 
 The integration of solar cells into various structures of the Robo Raven 
platform required new fabrication techniques to be developed.  The methods for 
developing these new structures have been established through this work using an 
additive manufacturing approach based on layered object manufacturing. The 
fabrication technique for wings with integrated solar cells was also used to integrate 
solar cells into the tail structure. This included both mechanical and electrical 
integration of the solar cells into these structures. Thus a FWAV that flew on both 
battery and solar power was created and flown.   
 This work also developed 5 new versions of the Robo Raven III vehicle.  The 
effects of hanging the wings design were clearly investigated.  Even though the larger 
version consumed more power, they lead to a design that has the ability to rely solely 
on solar power for flight.  Although Robo Raven III v5 was not actually made with 
the GaAs solar cells, it is the most impressive.  While still capable of flight, a swap of 
the copper used to simulate the GaAs cells for the actual solar cell will produce a 
FWAV that has the potential to fly completely on solar power.   
  146 
 
7.2 Characterization 
 
7.2.1 Characterization of Aerodynamic Forces 
 
 The forces produced by new wing designs was a major aspect of this work.  
Even though these loads have been measured in different ways in the past, the 
development of our test stand was novel.  Not only was the technology used better 
than what was used previously, but the flexibility to test different conditions while 
only making small adjustments to the test stand was new.   
 Previously, a single axis load transducer was used to measure the forces 
generated by a smaller FWAV [23,25,26].  This did not allow for simultaneous lift 
and thrust to be collected.  With the new test stand, a 6 DOF load transducer was used 
to simultaneously collect all of the loads and moments generated by the FWAV.  We 
focused on just lift and thrust; however, this test stand is capable of measuring much 
more.  As the work developed so did the test stand.  We were able to measure the 
actual flapping amplitude using an optical encoder and the power consumed by the 
vehicle.  We also made the test stand able to adjust the angle of the vehicle with the 
incoming air from the wind tunnel. This enabled actual flight conditions to be tested.  
This final test stand is a great tool that can be replicated in any lab and used to 
compare the behavior of different UAVs.   
7.2.2 Characterization of Deformation using DIC 
 
 This work is also one of the few times that 3D DIC has been implemented on 
a flapping wing.  This gave us valuable information as to what is occurring during the 
flapping process of the wing.  Obtaining the deformation of a flapping wing at 4 Hz is 
a difficult task that required knowledge of both the actual testing technique and the 
  147 
 
behavior of the FWAV.  Even though DIC has been used for a FWAV before [28], 
these are the largest wings yet to be observed. This provided new challenges that 
needed to be overcome.  For example the appropriate distance and location of the 
camera to the flapping wing needed to be found.  Also the correct amount of lighting 
for the indoor high speed cameras was also found.  Since the wings were so large 
typical speckling techniques could not be used and the wing needed to be hand 
painted.  These difficulties lead to an accurate and correct process that enable anyone 
to of similar wing size to use these techniques and find more information in wing 
deformation and performance. 
 One major contribution that was touched on before is the information 
extracted from the 3D DIC data.  A correlation between shear strain observed on the 
surface of the wing and thrust force generated was found.  A correlation between the 
biaxial strain observed on the surface of the wing and lift force generated was also 
found.  This new information in wing deformation allowed for a better wing design 
and lead to Robo Raven III v2. 
7.2.3 Electrical Performance Characterization 
 
  Since the advantages of implementing solar cells into a FWAV needed to be 
observed, this lead to the development of electrical testing procedures for Robo 
Raven III.  An endurance test was created that enables the new endurance of a solar 
powered FWAV to be determined.  This required new circuitry that enables the 
FWAV to operate using two power sources.  This circuitry involved a boost controller 
to increase the voltage of the solar cells to make it comparable to the voltage of the 
battery.  The test itself involved several aspect that cannot be controlled.  For example 
  148 
 
the weather had to be a clear day with no clouds.  Also, the FWAV needed to be held 
still at the correct azimuth to the sun to gain the most power possible.   
 The ability to recharge the battery was also tested for serval versions of Robo 
Raven III.  This required the development of a new circuit to safely charge the lithium 
polymer battery using the energy harvest by the solar cells.  Like the previous test, the 
weather also needs to be accommodating for these tests to take place.  Although these 
electrical tests were conducted for a FWAV, these tests can be conducted for any 
vehicle using solar power. 
7.2.4 Characterization of Self Sensing 
 
 An interesting finding in this work was how the solar cell was affected by the 
wing’s deformation.  This provides the wing with the ability to sense its own 
deformation using the power produced by the solar cell.  The correlation to shear 
strain and thrust provides available onboard information on vehicle performance. 
 Thus, it was determined that multifunctional solar cell wings may be capable 
of three functions:  
 (1) Lightweight and flexible structure to generate aerodynamic forces,  
 (2) Energy harvesting to extend operational time and autonomy,  
 (3) Sensing of an aerodynamic force associated with wing deformation. 
7.3  Modeling 
 
7.3.1  Multifunctional Modeling 
 
 A multifunctional performance model was developed.  This model broke the 
vehicle down in terms of mass and energy and allowed the increase in flight time to 
be predicted.  This model can be used for any UAV and is not limited to FWAVs. 
  149 
 
This means that the feasibility of introducing solar cells to any UAV can be 
determined before the project is started. 
7.3.2  Aerodynamic Modeling 
 
 The information gained from vehicle performance allowed for predictive 
aerodynamic models for lift and thrust to be generated.  The predictive models allow 
for changes in performance to be based off changes in design.  These models can be 
applied to FWAVs that are designed to fly laterally (usually with a wingspan of 
0.25m to 2.5).  One model was used to predict changes in wing performance.  The 
change in design causes a change in wing velocity that is used to predict the change in 
lift and thrust of the vehicle. The other predictive model was used to predict how 
changes in the tail would affect tail performance.  These models were based off the 
coefficient of lift and drag of a plane through air.   
 Finally by introducing stiffness as a component in wing design, the effects of 
integrating other materials to the wings can be predicted.  The model produced altered 
the previously mentioned wing lift and thrust models to compensate for global 
stiffness.   An equation for global stiffness allowed for a wing with different materials 
to obtain one global stiffness coefficient and be plugged into the lift and thrust model.  
This allowed for a more robust model that can be used not only for solar cells but 
different membrane materials.  In result from the fabrication, characterization and 
modeling work accomplished for the different version of Robo Raven III, the new 
flight times, recharge times, thrust forces, and lift forces were predicted and measured 
for each version created.  A table summarizing the work accomplished can be 
observed below. The total flight weight is the total weight of the FWAV and how 
  150 
 
much payload the vehicle can carry during flight.  By subtracting the vehicle’s 
weight, the payload for each Robo Raven III version can be found. 
Table 7.1: Summary of each version of Robo Raven III fabricated 
 Robo 
Raven III 
Version 1 
Robo 
Raven III 
Version 2 
Robo 
Raven III 
Version 3 
Robo 
Raven III 
Version 4 
Robo 
Raven III 
Version 5 
Predicted 
Increase in Flight 
Time (sec) 
29 42 54 68 
Infinite 
with 100% 
Sunlight 
Measured 
Increase in Flight 
Time (sec) 
27 38 42 50 
Was Not 
Measured 
Predicted 
Increase 
Recharge Time 
(min) 
74 41 27 20 0 
Measured 
Increase in 
Recharge Time 
(min) 
149 90 52 44 
Was Not 
Measured 
Predicted Total 
Flight Weight 
(grams) 
332 367 405 410 380 
Measured Total 
Flight Weight 
(grams) 
332 361 399 413 382 
Predicted 
Payload 
Including Battery 
(grams) 
41 53 35 29 40 
Measured 
Payload 
Including Battery 
(grams) 
41 47 29 32 42 
 
7.4 Future Work 
 
 Based off the progress and findings in this work, there are several directions 
that that can be taken for future work.  First, the actual GaAs solar cells can be 
integrated into the current Robo Raven III v5 design.  This would be a culmination of 
all of the concept and findings presented in this dissertation.  The results from the 
  151 
 
actual GaAs can also be compared to the predicted results obtained by the mock cells 
used.  To successfully power this FWAV on solar power, some investigation must be 
done on using a capacitor to control the flow of energy to the vehicle. From flapping 
the power supplied by the solar cells is expected to constantly change, so a capacitor 
is needed to supply constant power. 
 Another aspect that can be further explored is the solar sensing that was 
discovered in this work.  Actually using these solar cells as a sensor can provide real 
time feedback that would allow for autonomous flight corrections.   
 A valuable tool that can be developed from using more DIC data would be an 
aerodynamic model for FWAVs.  This model would incorporate the deformation and 
translation of the wing with the load produced due to the incoming air during flight.  
This involves simultaneous mechanical and fluid simulations. The results from the 
simulation can be compared to the measured DIC and load cell results.  This is a huge 
undertaking, but would be an extremely valuable tool for FWAVs. 
 One final direction that can be taken going further is to integrate flexible 
batteries to the wings.  With the development of thin flexible batteries, it is not far off 
to integrate these batteries similar to how the solar cells have already been integrated.  
This can be taken further and batteries and solar cells can both be integrated.  These 
are all extension of the work that has been accomplished in this dissertation. 
 
  
  152 
 
Bibliography 
 
[1] Gerdes JW, Gupta SK and Wilkerson S (2012) A Review of Bird-Inspired 
 Flapping Wing Miniature Air Vehicle Designs. ASME Journal of Mechanism 
 and Robotics. 4(2). 021003.1-021003.11. 
 
[2] Kumar V and Michael N, (2012) Opportunities and Challenges with 
 Autonomous Micro Aerial Vehicles. The International Journal of Robotics 
 Research. vol. 31,  pp. 1279-1291. DOI 10.1177/0278364912455954. 
 
[3] Sane, SP and Dickinson, MH (2002) The Aerodynamic Effects of Wing 
 Rotation and a Revised Quasi-Steady Model of Flapping Flight. Journal of 
 Experimental  Biology. 205: 1087–1096. 
 
[4] Yan J, Wood RJ, Avadhanula S, Sitti M and Fearing RS (2001) Towards 
 Flapping Wing Control for a Micromechanical Flying Insect. Proceedings 
 ICRA. IEEE  International Conference on Robotics and Automation. 2001, 
 4:3901-3908. 
 
[5] Ma KY, Chirarattananon P, Fuller SB and Wood RJ (2013) Controlled Flight 
 of a Biologically Inspired, Insect-Scale Robot. Science. 340(6132), 603–607. 
 
[6] Fenelon MAA and Furukawa T (2009) Design of an Active Flapping Wing 
 Mechanism and a Micro Aerial Vehicle Using a Rotary Actuator. Mechanism 
 and Machine Theory. 45(2):137-146. 
 
[7] Zdunich P, Bilyk D, MacMaster M, Loewen D, DeLaurier J, Kornbluh R, 
 Low T, Stanford S and Holeman D (2007) Development and testing of the 
 mentor flapping-wing micro air vehicle. Journal of Aircraft. 44(5):1701–1711. 
 
[8] Keennon M, et al. (2012) Development of the Nano Hummingbird: A Tailless 
 Flapping Wing Micro Air Vehicle. presented at 50th AIAA Aerospace 
 Sciences Meeting. Nashville, Tennessee. 
 
[9] Pornsin-Sirirak T, Tai Y, Ho C and Keennon M (2001) Microbat:  A Palm-
 Sized Electrically Powered Ornithopter. Proceedings of the NASA/JPL 
 Workshop on  Biomorphic Robotics. Pasadena, CA. 
 
[10] de Croon GCHE, de Clerq KME, Ruijsink R, Remes B and de Wagter C 
 (2009)  Design, Aerodynamics, and Vision-Based Control of the Delfly. 
 International  Journal of Micro Air Vehicles, 1(2):71-97. 
 
[11] Bejgerowski W, Ananthanarayanan A, Mueller D and Gupta SK (2009) 
 Integrated Product and Process Design for a Flapping Wing Drive-
 Mechanism. Journal of Mechanical Design. Vol. 50, pp.725-735. 
  153 
 
 
[12] Jones KD, Bradshaw CJ, Papadopoulos J and Platzer MF (2004) Improved 
 Performance and Control of Flapping-Wing Propelled Micro Air Vehicles. 
 Proceedings of the AIAA 42nd Aerospace Sciences Meeting and Exhibit. 
 Reno,  NV. 
 
[13] Anton, S. R., & Inman, D. J. (2008, March). Vibration energy harvesting for 
 unmanned aerial vehicles. In The 15th International Symposium on: Smart 
 Structures and Materials & Nondestructive Evaluation and Health Monitoring 
 (pp. 692824-692824). International Society for Optics and Photonics. 
 
[14] Shah, A., Torres, P., Tscharner, R., Wyrsch, N., & Keppner, H. (1999). 
 Photovoltaic technology: the case for thin-film solar cells. science, 285(5428), 
 692-698. 
 
[15] Aberle, A. G. (2009). Thin-film solar cells. Thin Solid Films, 517(17), 4706-
 4710 
 
[16] Antartis, D., & Chasiotis, I. (2014). Residual stress and mechanical property 
 measurements in amorphous Si photovoltaic thin films. Solar Energy, 105, 
 694-704. 
 
[17] Nemat-Nasser S, Plaistead T, Starr A and Amirkhizi A (2005)Multifunctional 
 Materials. Biomimetics: Biologically Inspired Technologies, Ed. Y. Bar-
 Cohen. CRC Press. 
 
[18] Bejgerowski W, Gupta SK and Bruck HA (2009) A Systematic Approach for 
 Designing Multifunctional Thermally Conducting Polymer Structures with 
 Embedded Actuators. Journal of Mechanical Design. 131(111009):1-8. 
 
[19] Thomas JP and Qidwai MA (2005) The Design and Application of 
 Multifunctional Structure-Battery Materials Systems. JOM. Vol 57 (3), pp. 
 18-24. 
 
[20] Thomas JP et al (2005) Multifunctional Structure-Plus-Power Concepts. 
 AIAA. 43rd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, 
 and Materials Conference, Denver, CO. 
 
[21] Yang LJ, Hsu CK, Ho JY and Feng CK (2007) Flapping Wings with Pvdf 
 Sensors to Modify the Aerodynamic Forces of a Micro Aerial Vehicle. 
 Sensors and Actuators A: Physical, Vol. 139 (1-2)pp. 95-103.. 
 
[22] Hsu CK, Ho JY, Feng GH, Shih HM and Yang LJ (2006) A Flapping MAV 
 with PVDF-Parylene Composite Skin. Proceedings of the Asia-Pacific 
 Conference of  Transducers and Micro-Nano Technology. 
 
  154 
 
[23] Wissman J, Perez-Rosado A, Edgerton A, Levi BM, Karakas ZN, Kujawski 
 M, Phillips A, Papavizas N, Fallon D, Bruck HA and Smela E (2013) New 
 Compliant Strain Gauges for Self-Sensing Dynamic Deformation of Flapping 
 Wings on Miniature Air Vehicles. Smart Materials and Structures. 22(8), 
 085031. 
 
[24] Mueller D, Gerdes JW and Gupta SK (2009) Incorporation of Passive Wing 
 Folding in Flapping Wing Miniature Air Vehicles. ASME Mechanism and 
 Robotics Conference. San Diego, CA. 
 
[25] Mueller D, Bruck HA and Gupta SK (2010) Measurement of Thrust and Lift 
 Forces Associated With Drag of Compliant Flapping Wing Air Micro Air 
 Vehicles Using a New Test Stand Design. Experimental Mechanics. Vol 50, 
 pp. 725-735. 
 
[26] Gerdes JW, Cellon KC, Bruck HA and Gupta SK (2013) Characterization of 
 the Mechanics of Compliant Wing Designs for Flapping-wing Miniature Air 
 Vehicles. Experimental Mechanics. DOI 10.1007/s11340-013-9779-5. 
 
[27] Hsu CK, Evans J, Vytla S and Huang P (2010) Development of Flapping 
 Wing Micro Air Vehicles - Design, CFD, Experiment and Actual Flight. 48th 
 AIAA  Aerospace Sciences Meeting. Orlando, FL. 
 
[28] Wu, P., Stanford, B., Bowman, W., Schwartz, A., & Ifju, P. (2009). Digital 
 Image  Correlation Techniques for Full-Field Displacement Measurements of 
 Micro Air Vehicle Flapping Wings. Experimental Techniques, 33(6), 53-58. 
 
[29] Pines DJ and Bohorquez F (2006) Challenges Facing Future Micro-Air-
 Vehicle Development. Journal of Aircraft. Vol 43 (2), pp. 290-305. 
 
[30] Cox A, Monopoli D, Cveticanin D, Goldfarb M and Garcia E (2002) The 
 Development of Elastodynamic Components for Piezoelectrically Actuated 
 Flapping Micro-Air Vehicles. Journal of Intelligent Material Systems and 
 Structures. 13(9):611-615. 
 
[31] Muijres FT, Johansson LC, Barfield R, Wolf M, Spedding GR and 
 Hedenstrom A (2008) Leading-Edge Vortex Improves Lift in Slow-Flying 
 Bats. Science. 319: 1250–1253. 
 
[32] Zhao L, Huang Q, Deng X and Sane S (2009) Aerodynamic Effects of 
 Flexibility in Flapping Wings. Interface. 7: 485–497. 
 
[33] Mahjoubi H and Byl K (2013) Trajectory Tracking in the Sagittal Plane: 
 Decoupled Lift/Thrust Control via Tunable Impedance Approach in Flapping-
 Wing MAVs. American Control Conference (ACC). pp. 4951-4956. 
 
  155 
 
[34] Arabagi V, Hines L and Sitti M (2012) Design and Manufacturing of a 
Controllable Miniature Flapping Wing Robotic Platform. The International 
Journal of Robotics Research. 0278364911434368. 
 
[35] Mueller TJ (2001) Fixed and Flapping Wing Aerodynamics for Micro Air 
Vehicle Applications. American Institute of Aeronautics and Astronautics. 
Reston, VA. 
 
[36] Tsai BJ and Fu YC (2009) Design and Aerodynamic Analysis of a Flapping-
Wing Micro Aerial Vehicle. Aerospace Science and Technology. 13(7), 383-
392. 
 
[37] Madangopal R, Khan Z and Agrawal S (2005) Biologically Inspired Design of 
Small Flapping Wing Bird Vehicles Using Four-Bar Mechanisms and Quasi-
Steady Aerodynamics. Journal of Mechanical Design. 127(4), 809-816. 
 
[38] Gerdes J, Holness A, Perez-Rosado A, Roberts L., Greisinger A, Barnett E, 
Kempny J, Lingam D, Yeh C-H, Bruck HA, and Gupta SK (2014) Robo Raven: 
A Flapping-Wing Air Vehicle with Highly Compliant and Independently 
Controlled Wings. Soft Robotics. 1(4), 275-288. 
 
[39] Perez-Rosado A, Griesinger AJG, Bruck HA, and Gupta SK (2014) 
Performance Characterization of Multifunctional Wings with Integrated Solar 
Cells for Miniature Air Vehicles. ASME 2014 International Design Engineering 
Technical and Computers and Information in Engineering Conference, Buffalo, 
NY. 
 
[40] Stanford, B, Ifju, P, Albertani, R, & Shyy, W (2008) Fixed membrane wings for 
micro air vehicles: Experimental characterization, numerical modeling, and 
tailoring. Progress in Aerospace Sciences. 44(4), 258-294. 
 
[41]  Tice, Brian P (1991). Unmanned Aerial Vehicles – The Force Multiplier of the 
1990s. Airpower Journal. 41-54. 
[42] Gerdes, J, Bruck H, and Gupta SK (2015) A Systematic Exploration of Wing 
Size on Flapping Wing Air Vehicle Performance. ASME Mechanism and 
Robotics Conference, August, Boston, MA. 
[43] Holness A, Bruck H, and Gupta SK (2015) Design of Propeller-Assisted 
Flapping Wing Air Vehicles for Enhanced Aerodynamic Performance. ASME 
Mechanism and Robotics Conference, August 2015, Boston, MA. 
  156 
 
 
[44] Bauhuis GJ, Mulder P, Haverkamp EJ, Huijben JCCM, Schermer JJ (2009) 
26.1% thin-film GaAs solar cell using epitaxial lift-off. Solar Energy Materials 
and Solar Cells. 93(9):1488-1491. 
 
[45] Lin Q, Huang H, Jin Y, Fu H, Chang P, Li D, Yao Y, Fan Z (2014) Flexible 
Photovoltaic Technologies. Journal of Materials Chemistry C. 2(7):1233-1247. 
 
[46] Roberts L, Bruck, HA and Gupta SK. (2014) Autonomous Loitering Control for 
a Flapping Wing Aerial Vehicle with Independent Wing Control. ASME 2014 
International Design Engineering Technical and Computers and Information in 
Engineering Conference, August, Buffalo, NY. 
 
[47] Perez-Rosado A, Gehlhar RD, Nolen S, Gupta SK, & Bruck HA (2015). 
Design, fabrication, and characterization of multifunctional wings to harvest 
solar energy in flapping wing air vehicles. Smart Materials and Structures, 
24(6), 065042. 
 
[48] Polhamus, EC (1966). A concept of the vortex lift of sharp-edge delta wings 
based on a leading-edge-suction analogy. National Aeronautics and Space 
Administration, 1-15. 
 
[49] Taylor GK, Nudds RL, & Thomas AL (2003) Flying and swimming animals 
cruise at a Strouhal number tuned for high power efficiency. Nature, 425(6959), 
707-711. 
 
[50] Perez-Rosado A, Bruck HA, Gupta SK (2015) Enhancing the Design of Solar-
Powered Flapping Wing Air Vehicles using Multifunctional Structural 
Components. ASME IDETC/CIE, Mechanisms and Robotics Conference, 
August 2015, Boston, MA. 
 
[51] Perez-Rosado A, Bruck H, Gupta SK (2015) Integrating Solar Cells into 
Flapping Wing Air Vehicles for Enhanced Flight Endurance. ASME. J. 
Mechanisms Robotics. doi:10.1115/1.4032411. 
 
[52] Oh, J., Lee, K., Hughes, T., Forrest, S., & Sarabandi, K. (2014) Flexible 
Antenna Integrated With an Epitaxial Lift-Off Solar Cell Array for Flapping-
Wing Robots. Antennas and Propagation, IEEE Transactions on, 62(8), 4356-
4361. 
  157 
 
[53] Videler, J.J. (2006) Avian Flight-Oxford Ornithology Series. OUP Oxford 
Press. Great Britain, UK. 
 
[54] Allain R. (2012) Why cant Humans Fly Like Birds. Wired Digital. Conde Nast 
Digital. January 2012.