AN IMPROVED LIGHTWEIGHT MICRO SCALE VEHICLE CAPABLE OF AERIAL AND TERRESTRIAL LOCOMOTION MATTHEW RYAN POLAKOWSKI

Transcription

1 AN IMPROVED LIGHTWEIGHT MICRO SCALE VEHICLE CAPABLE OF AERIAL AND TERRESTRIAL LOCOMOTION by MATTHEW RYAN POLAKOWSKI Submitted in partial fulfillment of the requirements For the degree of Master of Science Thesis Advisor: Dr. Roger Quinn Department of Mechanical and Aerospace Engineering CASE WESTERN RESERVE UNIVERSITY May, 2012

2 CASE WESTERN RESERVE UNIVERSITY SCHOOL OF GRADUATE STUDIES We hereby approve the thesis/dissertation of Matthew Ryan Polakowski candidate for the Master of Science degree. (signed) Dr. Roger Quinn (Chair of the Committee) Dr. Joe Prahl Dr. Paul Barnhart (date) 4/11/2012 ii

3 For My Parents: Maureen and Ray, who supported me both mentally and financially through my years at CWRU and put up with twenty-some years of stepping on Lego bricks as I developed my engineering judgment. iii

4 TABLE OF CONTENTS TABLE OF CONTENTS...1 LIST OF TABLES...4 LIST OF FIGURES...5 ACKNOWLEDGEMENTS...6 LIST OF ABBREVIATIONS...7 GLOSSARY...8 ABSTRACT BY CHAPTER 1 INTRODUCTION General Chapter Topics CHAPTER 2 PREVIOUS RESEARCH Micro Air Vehicles Ground Mobility Multi-Mode Vehicles Current CWRU MALV Design CHAPTER 3 SYSTEM REQUIREMENTS Mission Statement Concept of Operations System Modeling Functional Flow Block Diagram

5 3.3.2 Trade Studies Configuration selection Mass Budget Requirements CHAPTER 4 - CONCEPTUAL DESIGN Aerodynamics Propulsion Electronics and Control Wheel-legs Structures CHAPTER 5 - DETAILED DESIGN Propulsion Aerodynamics Electronics and Control Wheel-legs Structures CHAPTER 6 MANUFACTURING Wing Wheel-legs Frame

6 CHAPTER 7 - RESULTS AND DISCUSSION Verification and Validation Subsystem Tests Flight Tests Final Aircraft Specifications Final Mass Budget Requirements vs. Capabilities Comparison with Previous MALV Design CHAPTER 8 - CONCLUSIONS AND FUTURE WORK APPENDIX 1 MOTOR SIMULATION RESULTS APPENDIX 2 AERODYNAMIC FORCE BALANCE SPREADSHEET APPENDIX 3 AVL INPUT FILES APPENDIX 4 WHEEL-LEGS CONTROLLER SOURCE CODE APPENDIX 5 DRAWINGS PACKAGE APPENDIX 6 FLIGHT TEST PHOTOGRAPHS REFERENCES Parts Sources Design References

7 LIST OF TABLES Table 1: Wing loading trade study results Table 2: Typical wing loadings for R/C model aircraft Table 3: Typical power loadings for R/C model aircraft Table 4: Wheel-leg torque trade study results Table 5: Wheel-leg speed trade study results Table 6: Component category breakdown Table 7: Initial system requirements Table 8: Wing loading trade study with selected values Table 9: Motor, propeller, and battery options Table 10: Selected results from propulsion system simulations Table 11: MALV Mass Budget Table 12: Comparison of system requirements and capabilities Table 13: Comparison of MALV specifications with previous design

8 LIST OF FIGURES Figure 1: Example MAVs, AeroVironment Black Widow (left) and Wasp (right) Figure 2: Example ornithopter MAV, AeroVironment Hummingbird Figure 3: University of Florida flexible wing MAV Figure 4: CWRU stick insect robots, Robot I (left) and Robot II (right) Figure 5: CWRU cockroach robots, Robot III (left) and Robot V (right) Figure 6: CWRU early Whegs robots, Whegs I (left) and Whegs II (right) Figure 7: CWRU current Whegs robots, DAGSI Whegs (left) and Mini-Whegs (right).. 18 Figure 8: CWRU MALVs, early design (left) and developed design (right) Figure 9: Functional subsystem breakdown Figure 10: HS-130 airfoil shape and relevant curve data Figure 11: MALV geometry as modeled in AVL Figure 12: Root-locus plot from AVL for landing (1), cruise (2), and dash (3) Figure 13: Electrical system schematic (black: 0V, red: +V, yellow: data, blue: motor). 52 Figure 14: CAD rendering of MALV structure

9 ACKNOWLEDGEMENTS The CWRU Design/Build/Fly team who taught me how to build model airplanes and let me borrow space in their shop. Dr. Roger Quinn for believing in the potential of my design when it was still scribbles on a whiteboard and for advising me through the whole process. Dr. Rich Bachmann for the original proposal of a multi-mode vehicle and for accepting all of the times I laughed about his original design. Dr. Paul Barnhart and Dr. Joe Prahl for all the practice their courses gave me in working effectively without sleep and for agreeing to sit on my defense committee. Ken Moses and Brian Guzek for putting their thumbs on the line as test pilots. 6

10 LIST OF ABBREVIATIONS AVL Athena Vortex Lattice (software) BEC Battery Elimination Circuit CA CyanoAcrylate glue, also called Superglue CAD Computer Aided Design CNC Computer Numerically Controlled (machine) CWRU Case Western Reserve University EPS Expanded PolyStyrene (foam) ESC Electronic Speed Control (electronic throttle) FFBD Functional Flow Block Diagram GWS Grand Wing Servo (propeller manufacturer) LIPO Lithium Polymer battery MALV Micro Air-Land Vehicle MAV Micro Air Vehicle PWM Pulse Width Modulation R/C Radio Control, Radio Controlled 7

11 GLOSSARY Aspect Ratio For a rectangular wing, the ratio of span to chord; defines whether a wing is long and thin or short and fat Baby Orangutan Microcontroller Chord Combination of a programmable microcontroller and dual motor driver on a single board, includes 7 analog inputs and 16 digital input / output lines, used to interpret control signals to the wheel-leg motors Measure of the length of an airfoil, from leading (front) edge to trailing (back) edge Cubic wing loading Dihedral Dutch-roll Elevon A measure of how heavy an aircraft feels in flight, compares weight to volume based on wing area Aircraft roll-yaw coupling, a yawing motion produces a corresponding roll; also used to refer to the wing up-angle used as one method of creating this effect Dynamic instability mode caused by high dihedral, aircraft oscillates about axes at 90 degrees out of phase, wing tips move in elliptical paths Aerodynamic control surface that combines the functions of elevator and aileron Inrunner motor Style of electric motor that has a stationary housing containing a rotating shaft 8

12 Monocoque Structural design where a thin hollow shell is used as a major load carrying member Outrunner motor Planform Style of electric motor that has a stationary component inside the rotating housing The shape of an aircraft wing when viewed from the top or bottom R/C servo (control servo) Wheg Small DC motor with integrated position control and feedback, used to move control surfaces Robot mobility appendage that functions as a combination wheel and leg 9

13 An Improved Lightweight Micro Scale Vehicle Capable of Aerial and Terrestrial Locomotion Abstract by MATTHEW RYAN POLAKOWSKI Development of a new vehicle to fulfill the multi-mode locomotion reconnaissance mission proposed by the Biologically Inspired Robotics Laboratory at Case Western Reserve University is undertaken to improve the in flight handling characteristics of the current design without a loss in functionality. The vehicle is capable of both aerial and terrestrial locomotion through the incorporation of two wheel-legs at the front of a fixed wing micro air vehicle. The new vehicle weighs 7.8 ounces, less than half the weight of the previous design. The maximum dimension of the new vehicle is 25% smaller, with a wingspan of 12 inches. The reduction in weight and size allows a significant reduction in power consumption, with a corresponding increase in flight endurance. The new design has demonstrated its basic flight capabilities during a test flight sequence. 10

14 CHAPTER 1 INTRODUCTION 1.1 General The purpose of this thesis is to document the design of a new micro-scale vehicle capable of both air and ground locomotion to fulfill the multi-mode reconnaissance mission proposed by the Biologically Inspired Robotics Laboratory at Case Western Reserve University. A particular focus is the improvement of the in-flight handling and stability characteristics as compared with the previous design. Multiple modes of locomotion are prevalent in biological systems. Animals adapted for flight, such as birds and insects, can traverse long ranges efficiently, but are also required to traverse much shorter distances on the ground. Insects are especially adept at ground locomotion as well, with a gait known to be efficient and capable of maneuvering over rough terrain. Cockroaches are particularly efficient at running, reaching and climbing over obstacles. [18] While this motion is effective at traversing rough ground terrain, it is limited to a relatively low speed compared to body size. On the other hand, flight requires a greater energy investment, but allows travel at much higher speeds and over greater distances. However, there is a minimum distance over which flight is not practical. If this minimum distance is greater than the positioning accuracy required, a secondary mode of locomotion is required. A vehicle with the capability to select (autonomously or through external human control) the mode that is most energy efficient for the desired distance and terrain would have a greatly increased range over a ground vehicle, while still retaining the precision maneuvering capabilities. 11

15 This thesis takes the concept of a multi-mode vehicle through a series of design stages, beginning with the determination of system requirements and culminating in the testing of a prototype device. Each design stage develops more detail into the subsystems of the vehicle. The final vehicle design comprises a fixed wing aircraft that utilizes Whegs technology for ground mobility. The fixed wing aircraft is the most energy efficient method for mechanical flight over long range. The Whegs system is a highly capable drive mechanism that has been proven on many ground robots. 1.2 Chapter Topics Chapter 1, the current chapter, develops the purpose and inspiration for this project, and includes a list of chapter topics. Chapter 2 describes research into current systems for both ground and aerial locomotion to identify the key technologies that meet the goals of the project. It also identifies the inadequacies in the previous vehicles developed to accomplish multi-mode locomotion. Chapter 3 proposes a mission to be met by the multi-mode vehicle, and derives a series of system requirements from system models and trade studies. Chapter 4 discusses the conceptual design of the vehicle, including an analysis of commercially available components and a functional description of custom systems. Chapter 5 builds on the conceptual design to formulate a detailed design of the vehicle. This includes the generation of a complete CAD model. 12

16 Chapter 6 details the manufacturing procedures and methods selected to convert the detailed designs into physical parts and assemblies. Chapter 7 quantifies the subsystem and vehicle performance parameters, and validates these specifications against previous vehicles and the project requirements. Chapter 8 identifies potential areas for future improvements and developments of the design. 13

17 CHAPTER 2 PREVIOUS RESEARCH All engineering designs build on previous work. In this case, micro air vehicle technology and ground mobility systems are combined with current research in multimode vehicles to develop a new generation of micro air-land vehicle. 2.1 Micro Air Vehicles A small vehicle capable of flight provides an ideal sensor platform for many surveillance applications, both in military and civilian use. Flight provides a greater range and field of view than a similarly sized terrestrial vehicle. In 2000, DARPA began a program into Micro Air Vehicle research. The only requirement for the vehicles was a maximum dimension of 6 inches. A variety of applicants progressed to the prototype stage. [21] Figure 1: Example MAVs, AeroVironment Black Widow (left) and Wasp (right) [2] Black Widow, constructed by AeroVironment, is a 6 inch circular flying wing MAV with a flying weight of 50 grams. It has flight duration of 30 minutes and the capability to transmit color video. [10] AeroVironment also developed the Wasp series of MAVs. These larger aircraft with 12 to 24 inch wingspans had flight weights of 170 grams, but 14

18 longer durations up to nearly 2 hours. These flight durations are only possible for fixed wing propeller powered designs. Other designs submitted for the DARPA study included vertical flight capable designs, including helicopter-like quad-copters and flapping wing ornithopters. However, these designs are severely limited in flight duration due to the inherent inefficiency of vertical thrust flight. Figure 2: Example ornithopter MAV, AeroVironment Hummingbird [2] DARPA has also sponsored a student competition to develop micro air vehicles to complete simulated reconnaissance missions. The various student designs have progressed from 15+ inch spans down to around 6 inch spans to accomplish the mission. Examples of these vehicles can be found in the referenced design reports. The Notre Dame competition entry is described by Torres and Mueller. Spoerry and Wong describe the design and build of a micro air vehicle called Project Bidule. [20][19] A significant challenge faced in the design of aircraft at this scale is the instability compared to larger designs. This is due to a combination of atmospheric turbulence at the scale of the vehicle and lower moments of inertia at the smaller size. In collaboration with the University of Florida, the CWRU Biorobots Lab has developed a series of 15

19 flexible wing MAVs. The flexible wing allows some damping and passive control over this turbulence. However, the flexible wing design is not suitable for high wing loadings, as the airfoil shape will deform to a shape with a lower maximum lift coefficient, leading to a stall. [11][22] 2.2 Ground Mobility Figure 3: University of Florida flexible wing MAV [17] Mobility over uncontrolled surfaces is a great challenge facing ground robotics. Wheeled robots are efficient at moving over smooth surfaces, but fail at complex terrain. Legged robots require complex control systems and a large number of actuators. This makes them inefficient for smooth terrain, but capable of traversing more rugged surfaces. The CWRU Biorobotics Lab has developed several hexapod (6-legged) robots based on various insect models. Robot I and Robot II are based on the stick insect, while Robot III and Robot V are based on the cockroach. All showed great mobility over various surfaces, and the ability to adjust gait based on the surface. However, each of these robots required at least 2 actuators per leg. Robots I and II are electric powered, but Robots III and V require an off-board pneumatic system. [12] The weight of these actuators makes traditional legged locomotion impractical for a flying vehicle. 16

20 Figure 4: CWRU stick insect robots, Robot I (left) and Robot II (right) [7] Figure 5: CWRU cockroach robots, Robot III (left) and Robot V (right) [7] Whegs technology was developed by the CWRU Biorobotics Lab as a compromise between wheel simplicity and leg capability. This has been used to allow a robot to successfully traverse rough and uncontrolled terrain. Whegs are based on the cockroach s scrambling gait as it runs over rough terrain. Each wheel-leg only requires a single motor to operate, and in certain multi-wheel-leg designs power from a single motor can be transmitted to multiple wheel-legs. Mini-Whegs robots have four wheel-legs, are around 6 inches long, and are capable of moving over complex terrain at significant speed. [18][15] 17

21 Figure 6: CWRU early Whegs robots, Whegs I (left) and Whegs II (right) [8] Figure 7: CWRU current Whegs robots, DAGSI Whegs (left) and Mini-Whegs (right) [8] 2.3 Multi-Mode Vehicles Previous research has been undertaken into the utility of a surveillance vehicle capable of both long-range and short-range mobility. The ability to travel efficiently and quickly over a relatively long range would allow surveillance at a distance. The short range mobility would then allow the vehicle to reposition with accuracy for optimal reconnaissance data return. Such a design would greatly increase the reconnaissance capabilities of ground troops in combat or disaster search and rescue teams. [4][6] There have been several vehicle types proposed to accomplish this reconnaissance mission. Many of the vertical flight capable MAVs would be capable of the short range precision maneuvering. These designs have the benefit of using the same propulsion and control system for both long and short range movements. However, their limited flight time and high power consumption per maneuver would limit the effective mission range 18

22 and duration. A second design separates the short and long range systems into ground and aerial mobility systems. This is similar to many flying animals that have separate muscles and body parts for flying and walking. Insects in particular are equally adept at flying long distances and walking over rough, unpredictable surfaces. 2.4 Current CWRU MALV Design The CWRU Biorobotics Lab working in collaboration with the University of Florida has combined the flexible wing MAV design with the Whegs technology to achieve a multimode vehicle. The current design consists of a carbon fiber fuselage shell supporting two wheel-legs at the front end with the propeller between them. The flexible wing is constructed of a ripstop nylon sheet with a carbon fiber support framework. A U-shaped tail is supported on the rear of the fuselage to provide directional control through rudder and elevator control. Roll (aileron) control is not implemented due to the requirement for the flexible wing. [17] Figure 8: CWRU MALVs, early design (left) and developed design (right) [17] The original MAV had a 12 inch wingspan. However, the early wheel-leg drive system would be damaged during the high speed landings required by the high wing loading. 19

23 This required the design and installation of a slip clutch system to protect the gear trains. The additional weight of this system increased the landing speed even more, leading to further modifications to strengthen the carbon shell. Eventually, the wingspan was increased to 16 inches with a 300 Watt propulsion system in an attempt to lower the landing speed. This solution was successful, but pushed the limits of the wing aerodynamics with a total weight of over 16 ounces. It is not uncommon for the MAV to tip stall and crash as the pilot attempts to slow it down for landing requiring sophisticated carbon fiber strengthened fuselage and compliant landing gear. [17][3] The design process of this carbon fiber vehicle is documented in Richard Bachmann s dissertation. The final vehicle was capable of repeated landings with only an occasional broken propeller. However, its high weight decreased its ground maneuvering capabilities including maximum speed and obstacle height. Its high wing loading required a larger wing surface and high power motor that would increase the vehicle s visible and auditory signatures during a mission. [3] 20

24 CHAPTER 3 SYSTEM REQUIREMENTS The definition of specific system and subsystem requirements is necessary for the design of any complex system. The requirements allow subsystems to be designed separately from the main vehicle, and still interface properly during final integration. Defining requirements begins with a mission statement to focus the highest level requirements. A concept of operations is then generated to determine key design factors and requirements. A series of system decompositions and models analyze tradeoffs and engineering limitations. From these trade studies, requirements for each system and subsystem in the vehicle can be derived. 3.1 Mission Statement A small vehicle platform capable of multi-mode aerial and terrestrial locomotion would be a great asset in the development of reconnaissance sensor networks. Such a vehicle could be launched a significant distance from the target and have the range and speed to fly to the point of interest. The ground mobility would allow short range repositioning once at the target for better coverage. The system could then have the capability to fly back to the launch point or to a second target location some distance away. As defined in the original multi-mode vehicle proposal, the final vehicle must meet several requirements for mission utility and transportability. The vehicle should have an out and back flight range of 1 mile, with a total flight duration of 15 minutes. For transportation by an individual, the vehicle should have a wingspan less than 12 inches and a weight less than 1 pound. [4][6] 21

25 3.2 Concept of Operations Before defining a system model, a typical mission must be analyzed to determine key design issues and constraints. As with any aircraft mission, the long range flight phase of the mission involves takeoff, climb, cruise, descent, and landing. Takeoff is accomplished through a hand launch, which requires a sufficiently low stall speed. After takeoff, the aircraft climbs to its cruising altitude for the long cruise to the target. An efficient cruise requires a low drag airframe and an efficient propulsion system. The descent and landing phase must be controlled enough for a precision landing at the given target, which may be a building rooftop or other limited area. A slow descent and controlled landing puts strong limitations on the low speed stability and control of the vehicle. The short range relocation phase involves maneuvering with precision over unknown terrain. The vehicle must survive its landing and convert to a mode of operation that allows it to reposition over a range of tens of feet. This transition may involve the reconfiguration of the vehicle to allow passage through small gaps. After the short range maneuvering phase and actual reconnaissance phase are complete, the vehicle returns to a flight configuration for a second long range phase to either return to the launch point or a second surveillance location. The takeoff from the location can be accomplished through either direct thrust vectoring (vertical takeoff) or through a dive off of the roof edge. [4][6] 3.3 System Modeling Any vehicle is a complex system, but a multi-mode vehicle is made additionally complex due to the requirement for two separate drive systems. The functional flow block diagram 22

26 allows this complex system to be broken down into subsystems, and those subsystems into individual components. Each small piece is then simple enough to be modeled in a trade study to determine its limitations and requirements. A combination of different subsystems can then be proposed as a configuration. The selection of a general configuration drives the early conceptual design Functional Flow Block Diagram A functional flow block diagram is created to break down the complete system into a series of subsystems. Each of these subsystems fulfills a specific series of requirements, which flow upward into systems and the final vehicle. Micro Air Land Vehicle Propulsion Aerodynamics Structures Electronics and Control Ground Mobility Propeller Wing Wing Radio Receiver Motors Motor Tail Tail Speed Control Wheel-leg hub Battery Control Surfaces Fuselage Whegs Control Wheel-leg spokes Figure 9: Functional subsystem breakdown The major subsystems as shown are: propulsion, aerodynamics, structures, electronics and control, and ground mobility. Propulsion deals with the power system used during the flight phase of the mission. Aerodynamics incorporates the wing, tail, and control 23

27 surface sizing for stable flight. The structures subsystem sets requirements for the physical load capacities of each component. All onboard circuitry and power conversion components are covered under the electronics and control subsystem. The requirements for the short-range mobility phase are covered by the ground mobility subsystem and its components Trade Studies In order to define reasonable limits for interrelated requirements, a series of trade studies are performed. These simple models allow quick analysis of the effects of one variable on several others. Trade studies are performed on the wing, the propulsion system, and the wheel-legs to determine size and power limits. The wing trade study balances weight and wing area to calculate a wing loading that will determine the vehicle flight characteristics. A modification of the standard wing loading is the cubic wing loading, defined as the weight of the aircraft divided by the wing area to the 3/2 power. [20] Instead of comparing weight, which is a function of volume or length cubed, to wing area, which is a function of area or length squared, the cubic wing loading raises the wing area to the 3/2 power. This compares weight to a length cubed function, which means that it will scale more effectively with overall aircraft size. The following table shows the results of varying aircraft weight and wing area, where the wing loading values have units of ounces per cubic foot (a common unit for model aircraft). 24

28 MAXIMUM TAKEOFF WEIGHT (oz) WING AREA (in 2 ) Table 1: Wing loading trade study results A review of typical model aircraft shows that aircraft with similar flight performance tend to have the same cubic wing loading, regardless of scale. Some of the common categories of model aircraft are shown below along with their corresponding cubic wing loading and flight characteristics. [20] R/C AIRCRAFT CLASS CUBIC WING LOADING (oz/ft 3 ) FLIGHT CHARACTERISTICS Sailplane or Park Flyer 5 to 7 Extremely slow stall speed, easily blown around by winds Trainer 7 to 13 Low stall speed, docile handling, can fly in most wind conditions Sport 13 to 20 Faster flight, capable of most aerobatics, wind resistant Pylon Racer 20 to 25+ Extremely fast, may not be able to take off from ground, easily reaches stalled condition Table 2: Typical wing loadings for R/C model aircraft Selecting a maximum and minimum cubic wing loading involves balancing the desire for forgiving flight characteristics with the capability to penetrate the turbulent wind conditions that exist at MAV scale flight. 25

29 The propulsion trade study determines the required power and thrust for the aircraft to have the desired flight characteristics. These values are again given requirements based on typical model aircraft performance. The power loading (power per weight) of several common classes of model aircraft is shown below. [20] R/C AIRCRAFT CLASS POWER LOADING (W / lb) FLIGHT CHARACTERISTICS Hand launch park-flier Below 40 Cannot take off from ground, no aerobatic capabilities, very low maximum speed Sport trainer or easy flier 50 to 100 Ground take off, some aerobatic capability, reasonable flight speed Pattern flier or hotliner 100 to 150+ Ground take off, full aerobatic capability, high flight speed 3D aerobat 150 to 200+ Thrust can exceed weight, full aerobatic capabilities from hover Table 3: Typical power loadings for R/C model aircraft Selecting requirements for the power loading is a balance of having reserve power on tap and the extra weight of the high power propulsion components. This power can be used for either thrust or top speed, depending on the selected propeller and motor combination. A good balance between the two can again be found by using typical sport aircraft specifications. For an aircraft on the wing, the thrust should be between 25% and 50% of the maximum weight. For an aircraft on the prop, thrust can be considerably higher (between 50% and 150%+ of vehicle weight). However, flying on the prop is fairly inefficient, as the aircraft is behaving like a vertical takeoff design with a very small rotor diameter. [13] If the first two factors have been followed, the pitch speed of the propeller (zero thrust speed) should be between 2 and 3 times the stall speed of the vehicle. This is highly dependent on the aerodynamics of the design, but will provide a measurement to choose 26

30 WHEG DIAMETER (in) WHEG DIAMETER (in) between a high thrust, low speed propeller and a low thrust, high speed propeller that draw similar power. [20] The wheel-legs trade study determines the torque and speed requirements of the driving motors based on the size of the wheel-legs. The first study determines the torque (in ounce*inches) required to lift a given vehicle weight at a given wheel-leg diameter, assuming only one wheel-leg is lifting the vehicle. The second study determines the rotational speed (in RPM) required driving a given wheel-leg diameter at a given speed, assuming no slippage. WEIGHT (oz) Table 4: Wheel-leg torque trade study results SPEED (ft/s) Table 5: Wheel-leg speed trade study results Configuration selection The configuration of the vehicle involves how the major subsystems are combined into a single system. The wing and aerodynamics trade study shows the necessity to maximize 27

31 wing area for the lightest wing loading. This combined with the overall size limitations drives the planform to a rectangular, plank-style flying wing. This planform can be controlled with only two control surfaces (elevons) controlling pitch and roll. While this leaves no provision for active yaw control, it has been shown to be a sufficient control system for highly maneuverable radio-controlled aircraft. Simple vertical plates can be employed for passive yaw stability. Earlier studies on the Mini-Whegs robots show the greatest climbing capabilities when the wheel-legs are placed at the extreme front edge of the vehicle. [15] In order to minimize weight, only two wheel-legs will be used on the vehicle. The wheel-legs must also be placed at sufficient width to be able to turn the vehicle through differential (tank style) steering. Conversely, minimizing the width allows for the potential of squeezing the vehicle through a narrow gap if the wingspan can be reduced while on the ground. These two requirements drive a basic configuration with two wheel-legs set ahead of a rectangular wing. The area and structure between the wheel-legs and ahead of the main wing can be used for the flight motor and the electronics and control circuitry. The results of the propulsion trade study reveal that a single motor is the simplest and lightest option for the propulsion system. A tractor ( puller ) configuration is chosen, with the propeller fit between the wheel-legs. The propeller diameter is limited by the available ground clearance, dictated by the wheel-leg diameter Mass Budget One of the primary requirements for any aerospace system is mass because mass drives many of the other size and power requirements. Therefore, it is important to provide a 28

32 weight estimate for various subsystems and components as early in the design as practical. Components on the vehicle can be organized into four major categories: structure (the bare bones of the vehicle), flight hardware (the electronics necessary for propulsion and control), battery, and payload. The payload in this case is the ground mobility system, its associated control system, and any surveillance sensors such as a camera. This last category is slightly ambiguous; as its weight is driven by the leftover margin once the prototype vehicle is complete. A breakdown of the general components within each category is shown below. CATEGORY COMPONENTS Structure Wing Fuselage Tail fins Mounting points Flight Hardware Propulsion motor and prop Propulsion electronic throttle Radio receiver Control servos (2) Battery Propulsion battery Payload Wheel-legs (2) Driving motors (2) Control circuitry Additional sensors Table 6: Component category breakdown A review of commercially available electric model aircraft shows that structure and flight hardware weights are approximately equal, and their sum is equal to the battery weight. Since these models are not required to carry payload, it is assumed that the payload weight can be taken from the battery weight. This leads to an approximately equal distribution of weight over the four categories. Based on the wing loading trade study and the known configuration, an estimate can be made for overall weight. An average 3 inch diameter wheel-leg (the size of Mini-Whegs) results in a maximum wing chord of 9 inches, which limits the maximum weight to around 8 ounces for a sport wing loading. 29

33 This limits each component category to an initial weight estimate of 2 ounces. This estimate is refined as each component is developed through the design process. 3.4 Requirements Requirements beyond the original proposal are derived from the system models and trade studies. Each subsystem must adhere to its set requirements for mass and function if the entire vehicle system is to meet its mission requirements. Table 7 below summarizes the requirements for the subsystems. SYSTEM REQUIREMENT JUSTIFICATION Propulsion Thrust greater than 4 ounces 50% of takeoff weight Power between 50 and 75 watts Between 100 and 150 watts / lb Pitch speed between 2 and 3 times stall speed Sufficient thrust for cruise flight Battery weight less than 2 ounces Balanced system weights Propeller diameter less than 4 inches Ground clearance Aerodynamics Cubic wing loading less than 16 oz/ft 3 Desired flight characteristics Positive moment coefficient Static stability criterion Negative root locus points Dynamic stability criterion Ground Mobility Motor torque greater than 12 oz*in Climbing capability Ground speed greater than 3 ft/s Mobility requirement Overall Vehicle Total vehicle mass less than 8 ounces Balanced system weights Flight endurance of 15 minutes Mission definition Flight range of 1 mile Mission definition Table 7: Initial system requirements This preliminary list of requirements flows down to the conceptual and detailed design phases. More requirements are added as design decisions are made. Some subsystem requirements are also be changed or adjusted to maintain the overall vehicle and mission requirements. 30

34 MAXIMUM TAKEOFF WEIGHT (oz) CHAPTER 4 - CONCEPTUAL DESIGN The conceptual design phase of a project focuses on the development of a simple sketch of each system and subsystem. In this case, the major systems are defined to the point at which either a commercial solution can be selected from a small list, or the general dimensions and materials are chosen for the fabrication of custom components. 4.1 Aerodynamics The first step in the conceptual design of the aircraft is determining the proper relationship between wing area, weight, and cubic wing loading. The requirements state that the maximum wing loading is 16 oz/ft 3. Reviewing the wing trade study table and highlighting those combinations that fall below 16 oz/ft 3 results in the following table. WING AREA (in2) Table 8: Wing loading trade study with selected values The initial mass budget put the total weight of the aircraft at around 8 ounces. This gives an absolute minimum wing area of 96 in 2, while 108 in 2 would give a much more 31

35 comfortable margin and wing loading. Assuming the maximum allowed wingspan of 12 inches results in a minimum wing chord of 8 inches, with 9 inches being preferable. Conceptually, this low aspect ratio wing will be sensitive in several stability modes, in particular the Dutch-roll (roll yaw coupling) mode. In order to minimize this, vertical fins of appropriate area and shape will be added to the trailing edge of the wing. The elevon control surfaces will also be formed by the trailing edge, hinged at the appropriate point along the chord. 4.2 Propulsion Following this selection of wing size and shape and having an estimate for the maximum weight, the minimum required power and thrust can be estimated. The requirement for a minimum of 100 watts per pound and the aircraft weight of 8 ounces results in a minimum of a 50 watt propulsion system. The aircraft will fly much more comfortably with a higher power loading off 150 watts per pound, or 75 watts in the final vehicle. The thrust requirement is determined likewise. Using the absolute minimum, 2 ounces of thrust are required. Half of the total weight, or 4 ounces, would be much better. With the original configuration selection of a single motor, the ideal requirements of 75 watts and 4 ounces of thrust narrow the selection of commercial options significantly. The most critical decision that must be made early is the type of motor, inrunner or outrunner. Inrunner motors might be classified as typical electric motors; they have a cylindrical casing with a protruding shaft that spins. Outrunner motors are slightly different; in this case, the entire motor case spins around a central core that is mounted to the airframe. Because they are spinning a larger mass at a larger diameter, outrunner 32

36 motors will typically spin slower but with more torque than an inrunner for a given input of power. This makes them ideal for spinning larger propellers for efficient high thrust applications. However, the propeller diameter on the vehicle is limited by the size of the wheel-legs, so large (more than 5 inch) propellers are not possible. By limiting the search to available inrunners, only one brand and size of motor remained; the Neu Proton series. These 12 mm diameter motors are capable of handling up to 100 watts for short bursts, or 50 to 75 watts continuously. What is more, several different winds (speeds) are available, allowing optimization between a given input voltage and desired output rotational speed. Likewise, propeller selection is equally limited. Only the GWS Direct Drive series of propellers are available in the small (less than 4 inch) range required by the wheel-legs diameter. These propellers come in a variety of sizes and pitches, allowing optimization between thrust and speed. One of the largest limiting factors in micro air vehicle design is battery technology. Lithium polymer cells are a great improvement over nickel based cells, allowing much greater capacity for a given weight. Using the 2 ounce battery weight developed in the initial mass budget, approximately 10 watt hours of battery power can be carried. Assuming a worst case scenario of constant full throttle flight (75 watt draw), the vehicle would have an endurance of 8 minutes. Flying at partial power can extend this endurance significantly. To meet the 15 minute requirement set by the mission, the average power draw must be less than 40 watts. 33

37 4.3 Electronics and Control The electronics and control subsystem is comprised of three major components; the radio receiver, the electronic speed control, and the ground mobility control board. These components work to translate the pilot s input on the standard model airplane transmitter into signals usable by the various motors and servos that move the aircraft. First is the radio receiver. This converts the transmitter inputs to a Pulse Width Modulated (PWM) signal, where each control channel outputs a pulse ranging from 1 to 2 ms in length depending on the state of the channel. Second is the electronic speed control (ESC). This is a high current capable switching device that translates the receiver output pulse into the appropriate control signals for a three wire brushless motor. Third is the ground mobility control board. This works much like a dual ESC for the ground mobility motors. However, it also must have functionality to ensure that the ground mobility motors are not turning while the aircraft is in flight. The first major design decision is the selection of the appropriate radio band. Older remote control aircraft operate in either the 72 mhz band (mostly hobbyists) or the 27 mhz band (mostly toys). Recently, the 2.4 GHz band has been opened to hobbyists. Besides providing a more secure and stable communications protocol between pilot and aircraft, the 2.4 GHz band also provides benefits in the design of the aircraft. All radio receivers require a minimum of a quarter wavelength antenna to receive a signal efficiently. The 72 mhz band has a long wavelength, and the antennae for the receivers were typically around 3 feet long. This presents a design issue when integrating a 72 mhz radio into a small (12 inch) aircraft, as the antenna will tend to stream behind the aircraft causing drag and potential propeller interference. The 2.4 GHz band has a much shorter 34

38 wavelength, and the antennae are only around 1 inch long. This makes mounting the receiver and routing the antenna much easier in a small vehicle. For these reasons, a 2.4 GHz system is selected for this vehicle design. The second major decision regarding the radio selection is the number of control channels. At minimum the aircraft would require 5: throttle, left elevon, right elevon, left wheel-leg, right wheel-leg. However, a sixth channel is deemed necessary as the switching channel between flight and ground modes. It could also be potentially used for control of the additional sensors. The ESC selection is primarily driven by the propulsion system selection. One primary feature required by this design is a Battery Elimination Circuit (BEC) that converts the high voltage of the propulsion power battery to a more suitable voltage for driving the receiver and control surface servos. Similarly, the design or selection of the ground mobility control board is also driven by the specific selection of the ground mobility motors. However, it is guaranteed that this board will require some form of microcontroller to read in and time the receiver pulses and a motor driver to convert the low power output of the microcontroller to a suitable motor driving power. 4.4 Wheel-legs The ground mobility system makes up the majority of the payload of the micro air vehicle. Therefore, it is critical to minimize the system weight to minimize the penalty to flight characteristics. The system is comprised of two major components, the driving motors and the wheel-legs. The early choice of the wheel-legs for the ground mobility capability was made based on the high development level of the technology. The Mini-Whegs series of small robots has 35

39 demonstrated the superior capabilities of a three spoke wheel-leg design in maneuvering over obstacles. [18][15] The major design issue for the MALV wheel-legs is the minimization of weight while maintaining sufficient strength to survive a landing impact. Based on the previous design, plastic spoke wheel-legs would tend to crack under the landing impact. If the wheel-legs are manufactured from a flat sheet of material, the vertical area added to the front of the vehicle would tend to make the aerodynamics unstable. One way to alleviate the landing impact is through controlled structural compliance. Basically, the wheel-leg spokes are designed to bend slightly to absorb the impact and then spring back to their original shape. Spring tempered steel wire is ideal for this type of compliance, and has been used by many model aircraft designs as part of the landing gear. A potential second degree of compliance in the wheel-legs is rotational compliance. A mechanism could be developed to allow the motors to drive the wheel-legs up to the motor stall torque, but allow the wheel-legs to slip when driven beyond that torque. This would dissipate the high impulsive torque that happens as each wheel-leg impacts the ground during landing. The previous vehicle had a system with this functionality installed, but it significantly increased the weight of the ground mobility system. [17] The driving motors for the wheel-legs must provide enough torque to lift the front of the vehicle over obstacles and enough speed to move at a reasonable pace. Based on the trade study shown earlier, a 3 inch whell-leg (limited by 9 inch wing chord and 12 inch maximum dimension) on an 8 ounce vehicle would require 12 ounce inches of torque to 36

40 move. A search of available motors showed that the Pololu Micro Metal Gearmotors would be the best choice. This series of motors has 9 different gear ratios, but a common size and weight, and so can be easily interchanged if different characteristics are needed. Using the minimum 12 ounce inch torque requirement, the list of gear ratios can be narrowed down, and then a final motor selected based on the speed trade study. 4.5 Structures The conceptual design of the structure of the vehicle identifies the primary structural elements and selects the appropriate materials for each. Based on the conceptual designs of the other systems, the structural design of the vehicle has two major components, the wing and the fuselage. The structural design of the wing is focused on selection of materials and construction technique. The fuselage design requires a component layout as well as material and assembly decisions. There are three primary methods that could be used to construct a wing of this scale. A built up wing is constructed with ribs and spars similar to a full size aircraft. This construction can be very lightweight for the wing area. For this reason, most model aircraft are built with this technique. However, this construction is also susceptible to damage from impact. A second method is the current technique of carbon fiber with ripstop nylon. This creates a very durable wing, but is hard to produce repeatedly and requires a cambered plate airfoil shape with near zero thickness. The flexibility and compliance of the nylon may provide some passive stability, but the extremely thin airfoil is inefficient at high angles of attack (i.e. landing). The third construction method is a foam core composite. In this method, a lightweight beaded foam core is cut to the shape 37

41 of the required airfoil, and then covered with a layer of fiberglass or carbon fiber. This allows a thick wing that performs better at the high angles of attack. While this construction benefits the aerodynamics, it eliminates potential for simple wingspan reduction mechanisms. This design also results in a slightly heavier wing than a built up structure, but it can withstand a significant impact before requiring repair. This durability is important to a vehicle that may undergo rough handling during the proposed mission. Having developed a basic design for all of the other systems, conceptual design of the structure to connect them can begin. There are two basic options for the structure of a small radio controlled aircraft: monocoque shell or frame. The previous design is built with a monocoque shell, where all of the components are supported by an external shell that also doubles as an aerodynamic fairing. For this particular application, the placement of the wheel-legs far forward and wide on the fuselage means that a large volume is required to properly aerodynamically fair the wheel-legs driving motors. This increases the overall surface area of the shell, which becomes heavy when made thick and strong enough to support the landing loads. A framework structure allows the aerodynamic fairing to be made much lighter to only support aerodynamic loads, since the landing loads are transmitted directly to the dedicated frame. Based on the general configuration selected earlier, the frame is required to support the wheel-legs out in front of the wing at a wide spacing. Since the wheel-legs are the primary ground contact point during landing, this structure must support the entire inertial load of the vehicle as it lands. An obvious material choice to sustain this load is carbon fiber. Carbon fiber composite is extremely stiff and strong for its weight, and is ideal for high load structures. Pre-manufactured carbon rods and tubes are easily 38

42 assembled into a cross structure between the wheel-legs motors. Additional rods are used to transfer the propulsive thrust and wheel-legs loads from the cross bars into the main wing. This results in a T-shaped structure protruding from the front of the wing with end plates to mount the wheel-legs motors and propulsion motor to the carbon framework. 39

43 CHAPTER 5 - DETAILED DESIGN The detailed design phase of the vehicle involves the development of the conceptual design decisions into fully defined systems. At the completion of this phase, a fully dimensioned CAD model is created to enable virtual fit testing and easier final assembly of the vehicle. 5.1 Propulsion The conceptual design of the propulsion system limits the motor choices to the Proton series made by NeuMotors. Six winds are available in this line, ranging from 3200 RPM per volt up to RPM per volt. A search of GWS propellers less than 4 inches results in 5 different propellers. Table 9 shows the choices in motor and propeller that were combined with batteries ranging from 2 cells to 4 cells. MOTOR PROPELLER BATTERY CELLS (Voltage) Neu Proton 3200 GWS 2.5 x (7.4) Neu Proton 4000 GWS 2.5 x 1 3 (11.1) Neu Proton 5300 GWS 3 x 2 (trimmed to 2.5) 4 (14.8) Neu Proton 7300 GWS 3 x 2 Neu Proton 8300 GWS 4 x 2.5 (trimmed to 3.5) Neu Proton Table 9: Motor, propeller, and battery options As it is cost prohibitive and extremely time consuming to purchase every motor wind, propeller, and battery for an exhaustive test, a method of simulating the propulsion system is required. The Castle Creations ecalc software is known to be an accurate method for predicting propeller and motor performance. [9] The ecalc interface allows selecting battery, speed control, motor, and propeller from a predetermined database or the option to enter the specifications manually. As Castle is a distributor of NeuMotors, the Proton series are all available in the database. Likewise, all of the common propeller 40

44 brands are available, including GWS. For the initial run, both battery capacity and speed control current are oversized, which allows the motor to reach its full potential without being limited by the other elements. This initial run allows the estimation of maximum current draw for this range of system. Knowing this information is vital to the selection of the optimal battery capacity and throttle controller. Lithium chemistry batteries are rated by 3 main specifications: number of cells (voltage), capacity (size), and current capability (C rating). This last specification is an indication of how much current can be drawn from the battery without overheating it. It is usually specified as a multiple of the capacity. For example, a 1.5 amp hour capacity battery with a 20 C capability could source up to 30 amps of current without issue. C ratings are typically an indication of power density of the battery, with high C batteries being heavier for the same capacity. Based on the initial runs of the simulator, the maximum expected current draw is approximately 10 amps. The lowest C rated batteries available from ThunderPower RC are around 25C, which would require a 400 milliamp hour (mah) battery to supply the 10 amps. Any capacity larger than this will have a higher maximum current rating. Based on this and the requirement for a maximum 2 ounce battery weight, the maximum available capacity for 2, 3, and 4 cell packs is selected. For the 2 cell pack, a 1350 mah pack is just over the 2 ounce limit, but the additional capacity is worth the extra weight. The 3 cell pack has a capacity of 730 mah, and the 4 cell pack is 430 mah. The full results of the simulation are shown in Appendix 1, with any combination that would overload the motor dropped. These results are analyzed to determine which combination of motor and propeller would generate sufficient thrust and sufficient 41

45 endurance. Any combination with less than 4 ounces of thrust, less than 8 minutes of full throttle endurance, or power draw outside of 50 to 75 watts is eliminated. This elimination results in 5 possible combinations, shown in Table 10 below. MOTOR PROP CELLS mah CURRENT (A) THRUST (oz) TIME (min) POWER (W) x x x x x Table 10: Selected results from propulsion system simulations This list is now a manageable number of options to consider, and the final selection is made based on a variety of factors. The 4 cell combination is quickly eliminated due to the high voltage of the battery pack. This voltage would prevent the use of the BEC on most potential throttle controllers, and it would also require additional step-down to drive the wheel-legs motors. The 3.5 inch propeller combination is also eliminated due to the need to clip the diameter of a 4 inch propeller. This has a high potential for causing an imbalance in the propeller, which is very undesirable when the propeller is spinning at close to RPM. Of the three remaining systems, the 3x2 propeller combination was selected due to the higher propeller diameter. While the smaller propellers may have greater endurance or thrust, a larger portion of their thrust will be blocked by the wheellegs motors and other structure mounted close behind the propeller. If the additional endurance or thrust becomes necessary in the future, the motor and propeller can be easily switched, as all Proton motors have the same overall dimensions and weight. 5.2 Aerodynamics The detailed design of the vehicle aerodynamics begins with a more accurate analysis of the forces acting on the aircraft. There are 4 forces that act on any aircraft in flight: lift, 42

46 drag, thrust, and weight. The weight is estimated to be around 8 ounces. The other forces are strong functions of velocity. Lift and drag are calculated using the standard equations shown below, with the lift coefficient and parasitic drag coefficient as inputs. [13] The drag coefficient is the sum of 2 parts; parasite drag based on the shape of the vehicle and induced drag caused as a side effect of lift. Since parasite drag is a constant for a given airframe and induced drag is a function of lift coefficient and other design constants (shown below), the only variable input for the lift and drag equations is the lift coefficient. [13] Thrust is calculated as a linear relationship between static thrust at zero velocity and zero thrust at the propeller pitch speed. However, propellers with high pitch to diameter ratios may experience prop stall, which causes the thrust to limit at a lower value than static. The ecalc simulator takes this effect into account, and gives both the calculated static thrust and an estimated stall thrust. The estimated motor RPM is also used from ecalc along with the propeller pitch to calculate the pitch speed. This is defined as the forward velocity which causes the air to flow over the propeller blades at zero relative angle. Basically the wings that form the propeller blades stop producing lift at the point, and so thrust drops to zero. 43

47 The difference between lift and weight at a given velocity results in the net upward force at that velocity. Likewise, the difference between thrust and drag gives the net forward force. For a given lift coefficient, the aircraft will fly in the velocity regime where both the net forward and net upward forces are positive. Varying the lift coefficient and observing the velocity ranges at which the aircraft can no longer sustain flight results in an estimate for the flight envelope. For the aircraft dimensions defined in the conceptual design and suitable estimates for wing efficiency and drag, the aircraft can operate in a coefficient of lift range between 0.1 and 1.2. Appendix 2 contains snapshots of the spreadsheet used for these calculations at these extreme conditions, as well as at an average value of 0.6. Below 0.1, the aircraft can no longer fly fast enough to maintain lift; the decreasing thrust and increasing drag prevent it. This occurs around 75 ft/s, which is then defined as the aircraft s maximum level flight speed. Above 1.2, the increasing induced drag prevents the aircraft from developing enough forward force to maintain velocity. This occurs around 23 ft/s, but this may be below the stall speed for the aircraft. To further define this minimum velocity, the actual airfoil must be examined. Selecting an airfoil for a MAV in this size range is a difficult task. The size and flight speed of the vehicle tend to place it right in the transition region from laminar to turbulent airflow. This transition is defined by the Reynolds number, which is a nondimensional parameter relating inertial fluid forces to viscous forces defined below. [13] 44

48 In relation to an aircraft, the length L is defined as the chord length of the wing. Turbulent flow tends to begin between Reynolds numbers of to [5], where viscous forces are no longer effective at damping random disturbances. However, the wide chord wing used on this design pushes the operating Reynolds numbers up to the high end of this range and beyond. At the minimum speed of 25 ft/s, the Reynolds number is , which generates almost fully turbulent flow over the wing especially if small surface defects trip the boundary layer. At the high speed range, the Reynolds number is , which is well within the accepted turbulent flow regime. Being in the turbulent regime tends to make the behavior of the airfoil more predictable, so it is not as critical to selects a special low Reynolds number shape. This greatly increases the number of possible airfoil candidates for the vehicle. However, the selected planform reduces this list drastically. Since the design is an unswept flying wing, only certain airfoils called reflexed airfoils will provide stability. An aircraft with a normal tail relies on the tail to counteract the downward moment produced by the main wing airfoil. Even a swept flying wing can use an airfoil with a slight downward moment, as the trailing tips of the wing are far enough back to act as a tail. However, an unswept, or plank, flying wing does not have this tail surface and requires that the trailing edge of the airfoil be curved upwards. This provides a positive overall pitching moment, which is one of the criteria for static stability. Being limited to this class of positive moment airfoils narrowed the field of potential choices greatly. Searching the available literature on each of the various airfoils led to the selection of the HS-130 for the vehicle. This airfoil was originally designed for plank wing gliders, which means that it will have a high lift to drag ratio. The airfoil was also intended for large 45

49 trailing edge control surfaces, which will prevent surface blanking by the wing at high angles of attack. This airfoil and its relevant characteristics are shown in Figure 2. [1] Figure 10: HS-130 airfoil shape and relevant curve data As seen in the figure, the HS-130 airfoil is only capable of reaching a maximum lift coefficient of around 1.0 at low Reynolds numbers and around 1.1 at high Reynolds numbers. As these numbers are generated for an ideal two dimensional airfoil, the real airfoil will have a maximum lift coefficient of around 0.9. This generates a stall speed of around 25 ft/s, slightly higher than the previously predicted minimum speed. Throughout the design process, the aircraft was analyzed for stability by the Athena Vortex Lattice program written by Mark Drela. [14] AVL takes a series of files 46

50 describing the aircraft configuration and mass distribution as inputs and will output required control surface trim deflections and dynamic stability modes. Example input files for the design are provided in Appendix 3. To determine the moments of inertia required in the mass file, a detailed model of the design was maintained in Solidworks. After each major design iteration or component change, the AVL files were updated to reflect the change and the stability calculations were re-run. Early in the analysis it became apparent that the most critical stability mode for the vehicle is the Dutch-roll mode, where the aircraft rolls and yaws at 90 degrees out of phase. This is a common issue with any low aspect ratio (short, wide wing) design. The instability is caused by too little yaw stability or too strong roll-yaw coupling (dihedral). As the vehicle encounters turbulence that causes a yaw deflection, the dihedral effect causes it to begin rolling. The dynamic forces caused by the roll reverse the yaw movement, which leads to a roll in the opposite direction. [13] In this design, the limited moment arm for the vertical tail means that the yaw damping force is reduced. Any slight deviation will tend to grow until the aircraft crashes. This means that the vertical tail must be continuously resized (usually larger) to counteract this tendency. An initial design with only a single centerline tail was quickly abandoned, as the fin was required to be close to 6 inches tall to stabilize the aircraft. Tip plates are found to be more effective, but can quickly lose effectiveness if the base chord is too long. The final design limits the height of the fins to 3 inches above the wing surface. This will allow a consistent overall height when the vehicle is sitting on the ground to enable entrance into height confined spaces. The AVL model of the final aerodynamic surfaces is shown in Figure

51 Figure 11: MALV geometry as modeled in AVL It is found that the most dynamically stable configuration that does not require excessive control deflections for static stability places the center of gravity at approximately 1 inch behind the leading edge of the wing (11% chord). This seems far forward compared to typical aircraft (placed at 25% to 33% chord), until the rear section of the wing is considered as a tail. The lowers the effective chord of the wing to approximately 7 inches, placing the center of gravity at 28% of that chord. The AVL stability outputs for the final configuration are shown in Figure 12 below. The three run cases show the stability at a near stall (slow speed) condition, cruise velocity, and maximum speed. 48

52 Figure 12: Root-locus plot from AVL for landing (1), cruise (2), and dash (3) 5.3 Electronics and Control As defined in the conceptual design of the radio system, the radio receiver is a 2.4 GHz unit with at least 6 control channels. A search of commercial options reveals that both JR/Spektrum and Futaba manufacture 2.4 GHz systems. Spektrum was selected based on previous experience with that brand of radio. Several 6 channel options are available, 49

53 each with different unique features. The Parkflyer variation is designed with the lightest weight, but has a limited maximum range. The next lightest option is the glider version. This has full range capabilities, and also includes extended length antennae useful to this vehicle design. This is due to the fact that 2.4 GHz signals can be blocked or distorted by carbon fiber, as it is electrically conductive. The extended antennae allow the sensitive ends to be routed away from any interfering structure. In this case, the antennae will be routed along the bottom and top of the wings away from the frame. The selection of the motor, propeller, and battery determines the current required by the propulsion motor at less than 10 amps. The lightest commercial speed control available with a capacity of 10 amps is the Castle Creations Phoenix 10. The Phoenix 10 also has the capability to be customized for a given application through the CastleLink computer interface. This allows a user to adjust many of the timings and automatic limiters in the ESC. The control servos are selected from the range of extremely small size, high torque servos available from Dymond Modelsport. These servos are mainly intended for installation in radio controlled gliders, and so are designed to fit in very thin wings and fuselages but still offer the same torque and speed specifications available in larger servos. This small size will also prove convenient in this application to imbed the servos in the foam wing to minimize drag. The Dymond D-60 servo provides 24 oz-in of torque, which experience has shown to be sufficient for control surfaces in the size range of this vehicle. Each of the wheel-leg motors also requires a driving circuit. The radio receiver outputs a pulse signal every 20 milliseconds that varies from 1 to 2 milliseconds in length. Driving 50

54 a DC motor at variable speed requires a 0 to 100% duty cycle pulse width modulation signal. The smallest commercially available device to perform this conversion without additional user input is the Pololu TRex Jr, capable of driving 2 motors at 5 amps each. Used on the previous MALV design, this board is significantly above the power requirements for the Micro Gearmotors, which only draw a maximum of 1.6 amps. The board also has extra features that are not used in the current vehicle design. However, Pololu Robotics also manufactures a series of micro robot controller boards called Orangutan. These consist of a microcontroller and a small motor driver integrated circuit combined on a single circuit board. While not directly intended for this specific application, the digital input lines on the microcontroller can be used to read in the pulse lengths from the radio and perform the necessary mathematic conversion to a motor direction and speed. The motor driver chip on the board can provide up to 1 amp continuously, or 3 amps burst, to two independent DC motors. This is ideal for the selected wheel-legs motors, which will only rarely reach the stalled condition and draw above 1 amp. The final commented microcontroller code is shown in Appendix 4. Power for the radio receiver and control servos typically comes from the main propulsion battery through a voltage step-down circuit built into the electronic throttle called a Battery Elimination Circuit (BEC). The BEC in the Phoenix 10 is capable of supplying up to 2 amps at 5 volts, which is not enough to run the wheel-legs motors and controller as well as the flight controls. However, the Orangutan control board also contains a step down chip that can power the microprocessor while allowing the wheel-legs motors to run at the full propulsion battery voltage. A schematic of the vehicle electronics and control system is shown in Figure 13 below. 51

55 Figure 13: Electrical system schematic (black: 0V, red: +V, yellow: data, blue: motor) As shown in the schematic, the elevon servos are directly connected to the radio receiver. The mixing required to convert elevator and aileron commands is performed on the transmitter. The three data lines to the Baby Orangutan transmit signals for the wheellegs forward/reverse, left/right, and the switch between air and ground mode. The Baby Orangutan then performs the mixing on these signals to control the left and right wheellegs. The transmitter also contains the functions to disable the elevons when in ground mode and to map the rudder and flap receiver channels (connected to the Orangutan) to the elevator and aileron control stick. Keeping all major control functions on one control stick simplifies the pilot workload during a mission. 52

56 5.4 Wheel-legs During conceptual design, a search through available motors resulted in the Pololu Micro Metal Gearmotors series as the best driving motors for the wheel-legs. These motors are standard brushed DC motors connected to a gear train with reduction ratios ranging from 5:1 to 1000:1. The motors attached to the gearbox are available in High Power and Standard types, differing by about 3 times the amount of torque but 4.5 times the maximum current draw. According to specifications, the 50:1 High Power motor would meet the 12 oz-in requirement with a fair margin, providing up to 20 oz-in at stall. This gearing would also provide a fair amount of speed for the vehicle on the ground, spinning the wheel-leg at 600 RPM. Based on the early trade studies, this would result in a maximum ground speed of 6 ft/s. The wheel-legs themselves are constructed of a machined aluminum hub with bent spring steel spokes. The hub is composed of two pieces machined to hold the spokes sandwiched between them. This sandwich is then squeezed between two collets on the motor shaft such that the contact faces will slip under approximately 18 ounce inches of torque. This slippage prevents the wheel-leg from back driving the motor and potentially damaging gears under an impulsive load. The breakaway torque is high enough that it will not usually occur under the normal ground maneuvering. Because the coefficient of kinetic friction is lower than static friction, once the wheel-leg exceeds the breakaway torque it will continue spinning until it the torque drops to a lower value. Since the breakaway torque is lower than the motor driving torque and the reengagement torque is lower still, the gears should never encounter a situation under which they are loaded above the rated torque. 53

57 5.5 Structures The detailed design of the aircraft structure involved the sizing of all components and their connection points to withstand the maximum expected loads while minimizing material weight. As the proposed mission has the vehicle operating in an unplanned environment, many of these loads cannot be completely calculated. However, previous experience and research on model aircraft has allowed the estimation and scaling of various components. The low aspect ratio of the wing results in greatly reduced stresses as compared to a thinner, longer wing. Previous experience with an 18 inch span, 6 inch chord wing has demonstrated that 1 pound per cubic foot EPS foam is sufficient for the core and a single layer of 0.75 ounce per square yard fiberglass cloth is sufficient for the covering. [16] If additional reinforcement is needed around component holes, carbon fiber strands can be laid onto the wing before the fiberglass cloth. The carbon will add a degree of stiffness and tear resistance to the fiberglass. An important concern with embedding components in a fiberglass covered wing is the requirement to break the contiguous nature of the fiberglass cloth. This is especially important on the lower surface of the wing where the normal flight forces put the fiberglass in tension. This issue is mostly alleviated by the placement of all component recesses on the top surface of the wing. However, the connection between the frame and the wing will still occur on the lower surface. To maintain a continuous covering, a strip of fiberglass or carbon cloth will be laid over the grooves that receive the frame tubes. 54

58 Secure hinging of the control surfaces and mounting the control horns is necessary to minimize play in the control surfaces. The horns will be mounted in slots passing through the control surfaces. By using epoxy to affix them in place, they will bond with the fiberglass covering and become part of the surface structure. To hinge the surfaces, the fiberglass covering will be contiguous across the bottom of the wing and through the hinge line bevel. The fiberglass is thin enough that it will bend and flex at this joint instead of cracking. The vertical tail fins must be constructed from a thin material that is lightweight, yet strong enough to stand up under the aerodynamic loads. Birch plywood and G10 fiberglass board have approximately the same strength and weight. The final decision is made by the ease of laser cutting the plywood. A 1/16 th inch piece provides enough stiffness for the fins. The center fin is mounted in a similar manner to the control horns. The tip fins are attached with epoxy and a strip of fiberglass is laid around the bottom edge of the wing and fin for extra strength. Now that all of the components have been selected, the detailed design of the frame can be completed. From the conceptual design phase, the frame has the general shape of a T extending from the front of the wing. Provisions are made to mount one of the wheel-leg driving motors to each end of the cross, and to mount the main propulsion motor pointing forward at the intersection. These mounting points are under great stress as they transmit landing loads and propulsive loads from the motors to the structure. While it is possible to mold and make custom carbon fiber mounts for the motors, these are not easily repairable or replaceable in the event of a crash. Instead, a thin birch plywood plate is mounted to the ends of the carbon rods and the motor attached to the plate. Like the tail 55

59 fins, G10 fiberglass could be used instead, but is not due to manufacturing concerns. Laser cutting the wood ensures that the plates are accurate and easily repeatable, something not possible with hand molded and drilled carbon plates. The plates are held to the carbon frame tubes through 2 screws which pass through the plate and thread into a drop of thickened epoxy in the end of each tube. Once the epoxy hardens, the screw can be threaded in and out of the epoxy to remove and replace the plate. The other major connection point in the framework is the intersection of the main tubes with the cross tubes. Early designs used a built up plywood box or a drilled block of material to hold the tubes in place, but neither design could survive even small landing mishaps. The final design lashes the four intersections together using Kevlar thread soaked in CA. The Kevlar is strong enough to resist any landing force, and the CA hardens it to prevent any joint movement. The final connection between the wing and frame carries the entire load of the aircraft in flight. Extending the main carbon tubes nearly 75% of the wing chord not only stiffens the wing structure, but also provides a large gluing surface to effectively transfer the load without point loads developing. In order to facilitate the proper positioning and placement of all components, a detailed Solidworks CAD model of the vehicle is maintained. The CAD model is also instrumental to providing accurate masses and moments of inertia to the aerodynamics stability analysis. A complete CAD drawings package and purchased parts list is provided in Appendix 5. A CAD rendering of the frame and component placement is shown in Figure

60 Figure 14: CAD rendering of MALV structure 57

61 CHAPTER 6 MANUFACTURING The conversion of the detailed design CAD models into a physical device is a major stage of the design process. This process may also show issues with the design that are not apparent in the CAD model. These issues must be solved before the vehicle is completed. 6.1 Wing The wing begins as a 1 inch thick sheet of white beaded Styrofoam. This is then cut down using a 4 axis CNC hot wire cutter. This device consists of 2 X-Y frames that move in the pattern of the desired airfoil shape. One frame is positioned at each end of the Styrofoam blank. As the frames move, a thin piece of electrically heated wire is pulled through the foam in the shape of the airfoil. The wire melts a thin film of foam and cleanly cuts the airfoil. The extra outside foam is removed and the resulting wing core freed. The excess foam is saved to act as a female compression mold during covering. The control surfaces are formed from the trailing edge of the airfoil by separating the trailing edge from the wing and beveling to allow approximately 30 degrees of travel. The small strip of scrap that results from this bevel is saved to be used in the covering process. At this point, both the wing and the control surfaces are covered with a layer of fiberglass. The fiberglass is applied across the hinge line such that it will act as the surface hinge. The fiberglass is then saturated with West Systems epoxy resin. After wrapping in a layer of Teflon plastic, the wing is placed in the excess foam from the hot wire cutting. The beveled strip is placed in the space between the control surfaces and the wing. Placing the wing in this mold ensures that the fiberglass is evenly bonded to the 58

62 surface of the wing. After 24 hours of curing time, the wing is removed from the mold and the excess fiberglass is trimmed. The components that mount in the wing are laid out and traced in their respective locations, including the two channels to receive the frame. The holes for the components are then cut through the fiberglass and foam to the desired depth, and the components are inserted. The control servos are then covered with foam plugs that are sanded to match the contour of the airfoil. A thin slot is cut in the rear center of the wing to accept the plywood plate that becomes the vertical stabilizer. The wing tips are also prepared for the mounting of the tip plates by cutting thin sections of foam to the rear corners such that the tip plates will lie flush with the foam wing tips. Both the tip plates and central fin are attached with West Systems epoxy. A slot is cut in each control surface to receive a control horn. The horns are attached in the slots with epoxy, and then the control pushrods are routed to the servos through channels cut in the foam. 6.2 Wheel-legs The wheel-legs are manufactured in two major pieces, the hubs and the spokes. The hubs are machined from aluminum plate with a CNC mill. Semicircular grooves are cut in the face of the plate to receive the spoke wires. The grooves are made slightly shallow, so that the plates will grip the spokes tightly when compressed together. Three holes are also drilled and tapped to receive the 0-80 screws that will compress the hub plates. The spokes are cold bent from inch spring steel wire. The corners are rounded such that 59

63 the cold bending does not cause stress cracks in the steel. Detailed drawings of the wheellegs are in Appendix 5 in the drawings package. 6.3 Frame The frame begins with the laser cutting of the plywood mounting plates and trimming the carbon tubes to length. A drop of epoxy is pushed into the end of each tube before it is aligned with the precut screw holes in the plates. A 1-72 size screw is then inserted into the epoxied tube and the assembly is left in a jig to harden. When the epoxy has cured, the screws are broken loose, leaving epoxy threads inside the carbon tubes and allowing the plates to be removed and replaced as necessary. This procedure results in the two main frame sections (the long tubes and the short cross tubes). The two sections are attached using Kevlar thread to lash the tubes together at each cross point. The three motors can then be mounted to the frame using the appropriate 0-80 screws. The electronic throttle and the wheel-legs drivers are then wired to their respective motors and soldered into place. Shrink tube is fit over the electronic components to both secure them to the frame and to provide a small measure of waterproofing. The completed frame is then epoxied into the channels previously cut into the wing. The leads from the various components are routed to the receiver location and plugged into the radio receiver. These wires are actually run through tunnels cut in the foam core. This results in a cleaner installation with less drag. A brief power on test using the propulsion battery ensures that all connections are secure and correct. 60

64 CHAPTER 7 - RESULTS AND DISCUSSION The final stage of the design process is the testing of the system and the analysis of the results. This analysis includes both the verification that the vehicle performs within specifications and the comparison to previously designed similar vehicles. A detailed analysis of any component that fails to perform to requirements is also conducted. 7.1 Verification and Validation At each stage of the design and manufacturing process, the results must be verified that the requirements are met. Only once the subsystems have successfully met their individual requirements can they be assembled and tested as a complete vehicle Subsystem Tests During the design of each subsystem, small tests are performed to develop possible ideas or research solutions to issues. The development of the wheel-legs control board involved several tests to develop the details of the program code. Early examples used the integrated LED on the Baby Orangutan board for output to test the procedures used to read the radio control pulse lengths. Additional testing verified that the wheel-legs are attached to the proper channels of the radio receiver and that the proper inputs on the transmitter cause the proper response. A significant feature developed through a series of trial and error tests is the capability to switch between air and ground modes with a single switch on the radio transmitter. This switch activates a function in the transmitter to disable the elevons and sends a signal to the receiver (and the Baby Orangutan) to enable the wheel-legs motors. 61

65 The aerodynamic control servos are tested in a similar manner to the wheel-legs, both for correct connections and for proper response. The surfaces are also driven to their extreme positions to verify that there is no binding in the control rods or hinges. The propulsion system undergoes its own series of tests prior to integration in the vehicle. Thrust and RPM measurements verify the output of the simulations. The front motor mount plate is also tested for structural strength by clamping the main carbon tubes to a bench and running the motor to full thrust. The first test of this kind resulted in plate failure and lead to the implementation of a restraining band of heat shrink around the rear of the motor case. Further testing confirmed that this band could restrain the motor at full power Flight Tests Due to the inclement and unpredictable weather in Cleveland, early flight testing was limited to the verification of basic flight systems. The first flight tests consisted of low power extended glides with the motor running at partial throttle. These short flights allowed the pilot to properly trim the control surfaces and become accustomed to the response to control inputs. The initial response showed that the aircraft required larger control throws than predicted by AVL, at least at low flight speed. This was potentially a result of boundary layer effects, which AVL does not include in calculations. Once the aircraft was trimmed to glide straight and level, the next flight test attempted to increase the throttle and fly for a longer distance. It became apparent during these flights that the power system was not performing to the same level as observed in the initial tests. Subsequent testing of the installed system showed that the ESC was limiting the 62

66 throttle to maintain a 10 amp maximum current. The ecalc simulation had predicted a 9.6 amp draw that was also verified in the bench tests, which should not cause this limiting. Therefore, something was causing the motor to draw more power than expected. The major difference in the propulsion system between the flight tests and the bench tests was the inclusion of a prop saver during the flights. This adapter is firmly held onto the motor output shaft by a pair of set screws, which also serve as mounting points for a rubber ring looped over the propeller. This non-rigid propeller mounting allows the prop to move and flex to prevent breakage if it strikes the ground during landing. However, at the high RPM of this propulsion system (around RPM), the propeller balance becomes extremely critical. Bench testing with the prop saver in place showed that the propeller was becoming unstable around the 2/3 throttle point. This wobble then causes the motor to begin drawing more power which triggers the current limiting. As the limiter slows the motor speed, the prop recenters in the prop saver and the power draw drops off. The limiter stops and the motor revs up again to restart the cycle. This cycling added to the already sensitive handling to make the aircraft uncontrollable, and the power limiting prevented it from climbing. Simply removing the prop saver allowed the motor to rev to full power as in the bench tests. Beginning the flight tests without the prop saver immediately showed an improvement in performance. However, this came at the price of a broken propeller after almost any hard landing. This is because many of the early flight tests were conducted without the wheellegs installed. Not installing the wheel-legs reduced the potential impact on the wheellegs mounting plates and T-frame and allowed the testing to concentrate on the flight 63

67 systems. The wheel-legs were simulated with nylon blocks of the same weight press fit onto the motor shafts. After only 3 or 4 flights without the prop saver, the pilot was able to coax the aircraft around the test field in a complete lap with a controlled landing. However, the aircraft was very unstable in both pitch and roll and required delicate control inputs to prevent tumbling out of control. This instability did not appear to be the result of aerodynamic design as the aircraft would fly straight and level for short periods before a gust would upset it. The appearance of instability was caused by the extremely short response time to these gust events. This is a feature inherent to all small aircraft, caused by the small moments of inertia and aerodynamic damping forces compared to the gust loading. Increasing the aerodynamic damping is not an option for this design as it would involve increasing the wingspan or tail area beyond the size requirements. Increasing the moments of inertia is a possibility, by rearranging the internal components further from the centerline. However, this would require the wing to be strengthened to support the weight on the wingtips, as opposed to along the carbon rod spine. The final option to damp the gust response is the installation of rate gyros. These rotational accelerometers sense the aircraft motion around an axis and perturb the appropriate control servo signal to counteract that motion at a frequency greater than a pilot can perform the same corrections. In a typical aircraft, one servo is responsible for the motions about a single axis (roll, pitch, or yaw). However, in this design, the elevons each have an independent servo that responds to both roll and pitch. This means that one gyro is required for each elevon, set up to properly respond to both axes of rotation. By 64

68 mounting the gyro with its sensitive axis at an angle to the centerline of the aircraft, the gyro will register both pitch and roll. The ratio of response is determined by the angle, with angles closer to parallel responding more strongly to roll. As a first attempt, the gyros were installed at about 30 degrees from the centerline. This gives a strong roll response and a weaker pitch response, which correlated to the pilot s observations that the aircraft was more sensitive about the roll axis. On the first flight attempt with the gyros installed, the difference was immediately obvious. The aircraft exhibited no uncontrollable tendencies under normal flight conditions. The pilot was easily able to complete several laps of the test field and bring the aircraft in for a precision landing. Further flights allowed the pilot to expand the flight envelope to higher speeds. At high speed, it was noted that the aircraft would begin a rapid roll oscillation as the gyro gain became unstable. This is easily correctable through the incorporation of multi rate gyros or through further fine tuning of the gyro gain. By reducing the gain, the gyros become less effective at lower speeds but will not oscillate at the high speed. During the gyro stabilized flights, it was decided to install the wire wheel-legs on the aircraft and test the transition from flight to ground mobility. After each of the test laps and landings, the wheel-legs were found to be functional, if slightly ineffective at dragging the aircraft through the grass. This could be improved with the development of more effective wheel-leg spokes. Appendix 6 contains a photographic record of significant flight test events. 65

69 7.2 Final Aircraft Specifications The prototype vehicle has met or exceeded all major requirements set forth. The maximum dimension in both length and wingspan is 12 inches. At a flight-ready weight of 7.8 ounces, the aircraft has exceeded the 8 ounce maximum weight requirement. This weight also results in exceeding the cubic wing loading target of 16 oz/ft 3. The final calculated wing loading is 12.0 oz/ft 3. Static testing of the propulsion system confirms that it produces 5 ounces of thrust with power consumption approximately 75 Watts. Flight testing demonstrates that this is sufficient power for the aircraft to fly comfortably at around 2/3 throttle. Detailed flight envelope information is not available from the limited instrumentation employed during the flight testing. However, analysis of the flight videos has resulted in estimates for the speed range reached during testing. The movement of the aircraft is measured relative to stationary objects in the background of the image. This distance is then scaled by the known length of the aircraft and the frame rate of the camera to result in a flight speed. Averaging over multiple frames reduces the resolution error in the measurements. The tables of measured values and the conversion calculations are shown in Appendix 6 after the flight pictures. Calculations predicted a stall speed of approximately 25 ft/s, a cruise speed around 40 ft/s, and a top speed around 70 ft/s. The video analysis resulted in a landing speed of 19.3 ft/s, a cruise speed of 39.4 ft/s, and a top speed of 64.1 ft/s. The error in landing speed is likely due to the fact that the calculations did not factor in the possibility of a dynamic stall condition. It is likely that the high angle of attack seen in the landing footage caused 66

70 the wing to dynamically stall and the aircraft to begin flying on the thrust of the prop instead of wing lift. This is normally an undesirable condition, as it leads to control surface blanking and loss of control. However, this aircraft has large control surfaces that are partially in the propeller slipstream, allowing control to be maintained through the stall until landing. The lower observed top speed is likely due to the previously mentioned gyro oscillations not allowing the aircraft to reach its full potential. The selected video frames were prior to the onset of the oscillation, and thus the aircraft may not have been at its maximum level speed Final Mass Budget A final accounting of the components of the flight-ready MALV is shown in the table below. Note that the difference between the minimum flight equipment and the 8 ounce takeoff weight requirement is appropriated for the addition of mission specific sensors such as a camera and video transmitter. If required, the maximum takeoff weight could be extended to incorporate more sensors at the penalty of higher landing and flight speed. The prototype aircraft has successfully flown at a weight of 8.25 ounces. This extra weight is due to repairs performed after some of the early flight test crashes. Further flight testing can verify flight capabilities at higher weights. Other options for the inclusion of additional sensor payload include reductions in the battery weight at the expense of flight time or removal of the wheel-legs system. The latter option does remove the capability for ground mobility, but leads to the intriguing potential for a modular vehicle system. In such a system, a common interface between the wheel-legs crossbar and the main aircraft would be designed to allow the easy removal of 67

71 the wheel-legs and associated hardware to be replaced with a similar weight sensor package. By having the packages be a similar weight, the overall center of gravity of the aircraft will not change and so the general handling and stability characteristics will remain constant over the multiple configurations. CATEGORY COMPONENTS WEIGHT (oz) Structure Wing foam 0.54 Wing fiberglass 0.88 Tail fins 0.59 Carbon frame 0.24 Mounting plates 0.02 Screws 0.06 Epoxy 0.30 Flight Hardware Propulsion motor 0.52 Propulsion prop 0.04 Propulsion ESC 0.21 Radio receiver 0.19 Control servos (2) 0.42 Stability gyros (2) 0.30 Battery Propulsion battery 2.26 Payload Wheel-legs (2) 0.30 Driving motors (2) 0.68 Control board / wires 0.25 Additional sensors (opt) 0.0 (min) 0.2 (typ) 0.7 (max) TOTAL 7.8 (min) 8.0 (typ) 8.5 (max) Table 11: MALV Mass Budget Requirements vs. Capabilities The following table lists the requirements as defined in the early design phases alongside the specifications of the final flight capable vehicle. Each requirement is then assessed as met, exceeded, or failed. The failed requirements are given justification for why failing that particular requirement does not affect the success of the system as a whole. In particular, the actual pitch speed is much higher than that suggested by the rules of thumb, but the propulsion system still have sufficient static thrust to propel the aircraft. Also, the battery selected weighs slightly more than the 2 ounces allocated in the initial 68

72 weight budget, but the overall aircraft still falls under the 8 ounce requirement. The final 2 requirements for endurance and range have not been tested as of this writing, but all indications are that the aircraft will meet or exceed them. REQUIREMENT CAPABILITY RESULT Thrust greater than 4 ounces 5 ounces thrust Exceeded Power between 50 and 75 watts 70 Watts power Met Pitch speed between 2 and 3 times stall Pitch speed of 110 ft/s Failed, sufficient static thrust speed Battery weight less than 2 ounces Battery weight of 2.1 ounces Failed, met weight budget Propeller diameter less than 4 inches Prop diameter of 3 inches Exceeded Cubic wing loading less than 16 oz/ft oz/ft^3 wing loading Exceeded Positive moment coefficient Cm0 > 0 at 0 degrees AoA Met Negative root locus points All points negative Met Wheel-leg motor torque greater than Wheel-leg motors with 48 Exceeded 12 oz*in oz*in Ground speed greater than 3 ft/s 6 ft/s maximum ground speed Exceeded Total vehicle mass less than 8 ounces Vehicle mass of 7.8 ounces Exceeded Flight endurance of 15 minutes ~not tested~ ~not tested~ Flight range of 1 mile ~not tested~ ~not tested~ Table 12: Comparison of system requirements and capabilities 7.3 Comparison with Previous MALV Design Comparing the specifications of the new prototype with the current MALV shows a clear improvement in the critical aerodynamic performance parameters. The 50% weight reduction combined with the 10% increase in wing area results in a 60% reduction in cubic loading. This has been calculated to reduce the stall speed by approximately 35%. The reduction in weight and wing loading has also enabled a significant downsizing of the propulsion system by 70% power. Flight testing has shown that the new aircraft requires less pilot input to maintain stable flight, due to the inclusion of the stability gyros. Gyros have been incorporated on some of the past MALV designs, but incur a weight penalty and are not as effective since the previous design lacks aileron control. Table 13 below compares the specifications between the previous MALV design and the current prototype. [17] 69

73 PREVIOUS DESIGN PROTOTYPE SPECIFICATIONS Wing Span 16 inches 12 inches Wing Area 96 in in 2 Weight 16 oz 7.8 oz Cubic Wing Loading 29.4 oz/ft oz/ft 3 Electric Motor Power W 75 W Landing Speed (stall) 38 ft/s, hard tip stalls 25 ft/s (calculated), 20 ft/s (observed) Control Scheme R/E/T A/E/T Airfoil Shape Cambered flat plate Reflexed, 9% thick Planform Short coupled tail Plank flying wing Structure Monocoque carbon fiber shell Carbon framework, foam core wing Wing Folding? Possible, but too heavy No Table 13: Comparison of MALV specifications with previous design 70

74 CHAPTER 8 - CONCLUSIONS AND FUTURE WORK The goal of this project is to design and build a MALV with more favorable aerodynamic characteristics compared to the previous design while retaining the same overall functionality. The most significant improvement is the reduction in weight. This leads to lower power requirements, lighter wing loading, and lower stall speed. The weight reduction is accomplished through redesign of the wheel-legs driving circuitry and mechanics, decreasing the payload imposed on the aircraft portion of the vehicle, resulting in a lighter propulsion system. The structure is converted from a monocoque shell and flat wing to a frame structure and foam core wing. Although initial flight testing has revealed some control sensitivity, incorporation of rate damping gyros has rendered the aircraft stable and easy to fly. On landing, the lighter weight and lower speed reduce the impact energy that must be dissipated, reducing the possibility of vehicle damage during landing. Future improvements to the vehicle systems include the development of a wing retraction mechanism to minimize the vehicle footprint while on the ground. This is complicated by the solid foam core wing on the current design. Further work on the wheel-leg hub designs is also required. The current hub design protects the motors, but the wheel-legs sometimes continue to slip when in ground mobility mode after landing. This may be solved by further adjustment of the current design or an entirely new hub. Improvements to the wheel-leg spokes will also improve the ability for the vehicle to move on slippery or soft surfaces. These improvements are currently subjects of research in the Biorobotics Lab, and are primarily mechanical systems outside the main scope of this thesis. 71

75 APPENDIX 1 MOTOR SIMULATION RESULTS The following table shows the raw results from the ecalc propulsion simulation. Red highlights indicate a failure to meet specified criteria. The green rows indicate those combinations that met the initial criteria and were selected for further review. MOTOR PROP CELLS mah CURRENT (A) THRUST (oz) TIME (min) POWER (W) x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x

76 APPENDIX 2 AERODYNAMIC FORCE BALANCE SPREADSHEET The initial table shown contains the values held as fixed throughout the calculations. Static thrust (T0), stall thrust (Tstall), and RPM are all taken from the ecalc propulsion simulation. The 0.5 pound weight comes from the early requirements and conceptual design. The spreadsheet was run again once the final mass had been determined. CHORD 0.75 ft SPAN 1 ft AREA 0.75 ft 2 ASPECT RATIO OSWALD EFFICIENCY 0.9 CD T lb Tstall lb RPM RPM PITCH 2 in PITCH SPEED ft/s DENSITY lb/ft 3 WEIGHT 0.5 lb The following 3 tables show the spreadsheet outputs for lift coefficients of 0.1, 0.6, and 1.2 respectively. Velocities are given in feet per second and forces are given in pounds. The bold rows in the right hand columns indicate where both forward and upward forces are positive. This gives an estimate of achievable velocities at that coefficient of lift. 73

77 CL = 0.1 VELOCITY THRUST LIFT DRAG NET FORWARD NET UPWARD

78 CL = 0.6 VELOCITY THRUST LIFT DRAG NET FORWARD NET UPWARD

79 CL = 1.2 VELOCITY THRUST LIFT DRAG NET FORWARD NET UPWARD

80 APPENDIX 3 AVL INPUT FILES The first input file shown here is the geometry input. This describes the aircraft as a series of surfaces that are defined with some given camber (or flat). This file also defines the control surfaces as sections of the wing surface that AVL may change the angle to drive the overall moments to zero. MALV #X back, Y right, Z up #Mach #iysym izsym ixsym #Sref Cref Bref #Xref Yref Zref #C D #===================== SURFACE Wing YDUPLICATE 0.0 INDEX 1 SECTION AFILE hs130.dat CLAF 1.1 CONTROL elev CONTROL 77

81 ail SECTION AFILE hs130.dat CLAF 1.1 CONTROL elev CONTROL ail #======================= SURFACE Rudder INDEX 1 SECTION SECTION #======================= SURFACE Tip Plate YDUPLICATE 0.0 INDEX 1 SECTION SECTION

82 The second file is the mass definitions file. The definitions of Lunit, Munit, and Tunit allow the mass, length, and time to be input in different units than AVL will use to compute the final results. In this case, mass is input in pounds and length in inches. The gravitational acceleration and atmospheric density are also defined in this file. The last section defines the mass, position, and moments of inertia for the components of the aircraft. Since this aircraft was already CAD modeled, the overall mass and moments from Solidworks were used instead of breaking down into individual components and approximating. Lunit = ft Munit = slugs Tunit = 1.0 s g = 32.2!ft/s^2 rho = !slugs/ft^3 # mass x y z Ixx Iyy Izz Ixy Ixz Iyz

83 APPENDIX 4 WHEEL-LEGS CONTROLLER SOURCE CODE The source code for the Baby Orangutan microcontroller was written, compiled, and loaded through the AVR Studio 4 development environment with the gcc-avr compiler package. The referenced orangutan.h library is available through Pololu Robotics (the manufacturer) free of charge. //MotorDriver.c //read in 3 standard RC servo pulses (enable, translation, rotation) //output to 2 standard PWM motors (left, right) #include <pololu/orangutan.h> const unsigned char pulseinpins[] = {IO_B0, IO_B1, IO_B2}; int main() { set_digital_input(io_b0, HIGH_IMPEDANCE); set_digital_input(io_b1, HIGH_IMPEDANCE); set_digital_input(io_b2, HIGH_IMPEDANCE); pulse_in_start(pulseinpins, 3); signed int offset = 32; signed int pulserun; signed int pulsemov; signed int pulserot; signed int whegl; signed int whegr; while (1) { pulserun = pulse_to_microseconds(get_last_high_pulse(1)); pulsemov = pulse_to_microseconds(get_last_high_pulse(2)); pulserot = pulse_to_microseconds(get_last_high_pulse(0)); if (pulserun > 1500) { red_led(1); whegl = pulsemov + pulserot ; //results in to 1000 whegr = pulsemov - pulserot; whegl = whegl*(255.0/1000.0); //results in -255 to 255 whegr = whegr*(255.0/1000.0); 80

84 } return 1; } } else { } if (whegl < offset && whegl > -offset) //set deadband { whegl = 0; } if (whegr < offset && whegr > -offset) { whegr = 0; } set_m1_speed(-whegl); //assign motor speeds set_m2_speed(-whegr); red_led(0); set_m1_speed(0); set_m2_speed(0); 81

85 APPENDIX 5 DRAWINGS PACKAGE The included drawings package is not intended to serve as a complete manufacturing package, but as an overall reference to aid in comprehension of textual descriptions of components. Some dimensions may vary slightly from the prototype due to unforeseen or unrecorded adjustments during manufacturing. Drawing 1: Overall MALV component layout Drawing 2: Detail of T-frame component layout Drawing 3: MALV 3-view with overall dimensions (all dimensions in inches) Drawing 4: Dimensioned detail of laser cut parts (dimensions in inches) Drawing 5: Dimensioned wheel-legs components (dimensions in inches)

86 Drawing 1: Overall MALV component layout 83

87 Drawing 2: Detail of T-frame component layout 84

88 Drawing 3: MALV 3-view with overall dimensions (all dimensions in inches) 85

89 Drawing 4: Dimensioned detail of laser cut parts (dimensions in inches) 86

90 Drawing 5: Dimensioned wheel-legs components (dimensions in inches) 87

91 The following table documents the purchased parts for the final MALV design. MANUFACTURER PART DESCRIPTION UNIT NUM TOTAL ThunderPower G6 Prolite 2s 1350mAh LiPo battery $ $22.99 Castle Creations Phoenix 10 brushless ESC $ $59.95 Neu Motors Proton brushless inrunner, 7300 kv $ $39.95 GWS 2 blade direct drive 3x2 propeller $ $0.75 Spektrum RC AR6255 six channel 2.4GHz reciever $ $79.99 GWS PG-03 micro piezo gyro $ $73.90 Dymond RC D60 micro servo $ $39.90 Pololu Baby Orangutan B-328 controller $ $19.95 Pololu Micro metal gearmotors HP, 50:1 ratio $ $31.90 TOTAL $

92 APPENDIX 6 FLIGHT TEST PHOTOGRAPHS Photo 1: Preflight checks on MALV, notice wheel-legs replaced with equivalent weight plastic 89

93 Photo 2: MALV with wheel-legs maneuvering on grass surface 90

94 Photo 3: A sequence showing typical hand launch of MALV 91

95 Photo 4: Partial sequence of photos used for dash speed measurements 92

96 Photo 5: Partial sequence of photos used for cruise speed measurements 93

97 Photo 6: Partial sequence of photos used for landing speed measurements 94