The document summarizes research being conducted on semi-autonomous hand-launched rotary-wing unmanned air vehicles (RUAVs) at Penn State. The research aims to design avionics systems using low-cost processors and sensors to enable reliable semi-autonomous control of small electric-powered quad-rotor RUAVs. Several processing and sensor options are being evaluated. Preliminary flight testing is being conducted using a PC-104 system on a Draganflyer III airframe. A custom-designed autopilot board is also being developed and tested. The researchers are taking an incremental approach to control design, starting with stability augmentation and moving towards semi-autonomous and autonomous capabilities.

1. PI’s : Prof. Lyle Long and Prof. Joseph F. Horn Tel: (814) 865-1172 and (814) 865 6434 Email: lnl@psu.edu and [email_address] Graduate Students: Wei Guo, Ph.D. Candidate Scott Hanford, M.S. Candidate Project PS 8 Semi-Autonomous Hand-Launched Rotary-Wing Unmanned Air Vehicles 2004 RCOE Program Review May 4, 2004

5. Processors and Software We are currently evaluating all of these Others: Motorola, HandyBoard, Atmel,… Very fast, can use c, light weight, inexpensive, … 1000 Basic, C, Java 1000 Yes Yes 128 MB 266 MHz JumpTec ( www.adastra.com ) (Intel PC Processor) PC-104 200 Basic, C, Java 1000 Yes Yes 1 GB 1000 MHz VIA MicroATX ( www.via.com.tw ) VIA 140 Basic 80? No Yes 34 KB 100 KIPS Basic Micro (www.basicmicro.com) (Hitachi 3664) BASIC Atom Pro 89 Java (subset) 60 No No 64 KB 8 KIPS Parallax (www.parallaxinc.com/) (Ubicom SX48AC) Javelin Stamp Yes Yes Floating Point Math? 100 Java 250 Yes 1 MB 40 MHz Dallas Semiconductor (www.ibutton.com/TINI) TINI Board 100 Basic or C 80 No 1800 KB 40 MIPS MicroChip (www.microchip.com) ( PIC18F452) PIC Cost with Board ($) Software Power (mA) Ether-net Memory Speed Manufacturer (website) Processor

6. Processors & Boards PIC Development Board PC-104 Micro ATX 7 inches 3.5 in. 4 in. (includes breadboard for sensors) Too heavy, complicated & power hungry

7. Sensors Analog Devices MEMS sensors ADXL 202/210 2-Axis Accelerometers Range: ±2g or ±10g RMS Noise : < 0.002 g for 10 Hz Bandwidth Option Power : 0.6 mA @ 3-5 VDC Size: Less than 0.5 gram, 0.05 cm 3 ADXRS 150/300 Rate Gyros Range: ±150 or ±300 °/sec Noise : 0.05 °/sec/Hz 1/2 Power: 6.0 mA @ 5 VDC Size: Less than 0.5 gram, 0.15 cm 3 Crista IMU Uses Analog Devices MEMS sensors, with 3 gyros and 3 accelerometers integrated into a single unit with serial interface. Unit weighs 37 grams with enclosure. This is a relatively expensive item. Very easy to use and integrate. Honeywell HMR 3100 Digital Compass One of the smallest and cheapest units available. Size: < 1.5 grams Power: 0.2mA @ 3 VDC RMS Accuracy : < 5 deg Novatel SuperStar II GPS Receiver At 22 grams, one of the smallest and lightest units available. Size: 22 grams Power: < 0.5 W Accuracy: < 5 m CEP (< 1 m CEP DGPS) Currently looking into SONAR altimeters All of these devices are currently being tested. Can survive 1000 g shock

8. Summary of Quad-Rotor Systems Currently Under Development 1. PC/104 Based System PC system is easy to program and to integrate with other hardware. Also uses Crista IMU, and wireless network adapter for communication. Too heavy and power consuming but will be use for rapid prototyping of control laws. 3. BasicAtom System 2. PIC Microcontroller System Very good performance relative to weight, power consumption, and cost. Programmable in C. Requires more electronics expertise to hardware. Lightweight, low power consumption. Program in BASIC. Can perform floating point math. Already integrated with MEMS sensors and motors. 4. Javelin Stamp System Lightweight, low power consumption. Program in JAVA. Limited to integer math. May not be feasible to implement control laws. 5. Custom-Designed System Custom-designed board integrates processor and all sensors and servos. Optimal in terms of weight and power. Requires EE expertise.

9. Draganflyer III Photo from www.rctoys.com SuperStar II GPS Receiver (22 grams) Remove existing electronics Crista IMU (37 grams) Intel 166 MHz Pentium MMX PC/104+ (110 grams) HMR3100 Digital Compass (1.5 grams) NetGear Wireless Network Adapter (20 grams) PC/104 is relatively heavy and power consuming PC Based Avionics

11. Quad Rotor Draganflyer III Photo from www.rctoys.com Remove Existing Electronics PIC Microcontroller or other lightweight microprocessor 4-channel receiver Wireless Video Camera MEMS-based ADXRS150 Gyro CMC GPS Receiver MEMS-based ADXL202E Accelerometer PIC Microcontroller Avionics HMR3100 Digital Compass

12. BasicAtom Test Platform Batteries Electronic Speed Control 280 size Motor Accelerometer Gyroscope BasicAtom Processor Serial PWM Controller 9 inch prop

13. Close-up of Processor and Sensor Boards Accelerometer Gyroscope Hitachi 3664 Processor 1 inch

14. Accelerometer Output X accel (in g's) = 0.000000 X accel (in g's) = 0.035693 X accel (in g's) = 0.014277 X accel (in g's) = 0.963735 X accel (in g's) = -0.513993 X accel (in g's) = 0.064248 X accel (in g's) = 0.021416 Acceleration in the positive x-direction

15. Gyroscope Output Z ang velocity (degrees / sec) = 0.000000 Z ang velocity (degrees / sec) = 0.000000 Z ang velocity (degrees / sec) = 14.062715 Z ang velocity (degrees / sec) = 47.266346 Z ang velocity (degrees / sec) = 28.516060 Z ang velocity (degrees / sec) = 4.296940 Z ang velocity (degrees / sec) = -1.171893 Z ang velocity (degrees / sec) = -18.359653 Z ang velocity (degrees / sec) = -26.953535 Z ang velocity (degrees / sec) = -37.109940 Z ang velocity (degrees / sec) = -37.109940 Z ang velocity (degrees / sec) = -36.328678 Clockwise Rotation Counter-Clockwise Rotation

16. PSU Custom Designed Autopilot 3 EE PhD students from Long’s UAV course developed a custom-designed avionics board using the MEMS-based sensors. Board was designed, manufactured, and is now undergoing testing and programming. Should be completed Summer 2004. Quad-rotor helicopter Gyros Atmel Processor Accelerometer Motor controllers

19. Control Law Design SIMULINK Diagram of Simple PID Controller

21. Autonomous UAV’s are vulnerable to the failure or degradation of sensors Incorporate multiple inexpensive accelerometers and take advantage of kinematical coupling to derive rate and attitude information Use an Extended Kalman Filter (EKF) to fuse sensor data Fault-Tolerant UAV Flight Control Long term objective FCC Accelerometers Flight Control Computer

22. New UAV Course PSU Funded, Taught by Prof. Long, www.personal.psu.edu/lnl/uav Fall 2003. 31 students Every student built and flew airplanes. Guest lectures on UAVs. 1 credit. Spring 2004. 28 students Students worked in teams of 5 to build large 80 in. span aircraft. Installed wireless video cameras, onboard flight data recorders, performed flight tests. 2 credits.

23. Onboard Wireless Video Camera View of area around flying field. At approximately 400 feet altitude. Cars Us Landing strip

24. Onboard Flight Data Recorder (Crash in UAV course, April, 2004) Altitude (ft) Speed (MPH) Onboard “blackbox” used for flight testing, but also contained crash data. (in high-speed dive the elevator failed and pilot lost control) Pilot Input Elevator lost, but pilot still trying… time, seconds