Construction, Modeling and Automatic Control of a UAV Helicopter

Transcription

1 Construction, Modeling and Automatic Control of a UAV Helicopter BEN M. CHENHEN EN M. C Department of Electrical and Computer Engineering National University of Singapore 1

2 Outline of This Presentation 1. Brief introduction to unmanned aerial vehicle (UAV) 2. Hardware construction 3. Ground station software platform 4. Modeling the dynamics of the UAV helicopter 5. Automatic controller design 6. Demonstration of actual test flights 7. Continuing work 2

3 Introduction to UAV Arouse great interest in both military and civil applications. Predator Dragonfly UAV AVATAR UAV helicopter A hot topic in academic community. Many universities have developed their own UAV test beds. 3

4 Hardware Construction HeLion,, Home Made UAV Helicopter Bare Helicopter Command Control Signal Real-time Data Onboard System Measurement Operation Ground Station RC Joystick 4

5 Hardware Bare Helicopter Raptor 90 SE Product of Thunder Tiger Length m (55.50") Width m (7.50") Height m (18.25") Weight - 4.8kg (10.5 lbs) Diameter of Main Rotor m (64.75") Diameter of Tail Rotor m (10.25") 5

6 Hardware Onboard Computer System PC-104 computer stack: 1). Main processing board 2). Serial communication board 3). Data acquisition board 4). DC-to to-dc converter board 6

7 Hardware Inertial Measurement Unit (IMU) IMU (Inertial Measurement Unit): Most important sensor in HeLion, collecting in-flight data including: 1). Accelerations & velocities 2). Rotating angles 3). Heading angles 4). 3D-GPS signals 7

8 Software System Structure Onboard System Ground Station System 8

9 Onboard Software Structure An Onboard System Execution Cycle: Processing IMU Signals Data Logging Data Acquisition Control Law Calculation Communications with Ground Controlling Servos 9

10 Ground Station Software Interface Connection Button Graphical Display Optional Data Display 3D View Online Data Display Command Window Gauge View 10

11 Ground Station Software 3D View 3D View Real Fly 11

12 Modeling Data Collection Data Collection Procedure: 1). Chirp-like signal issued in single channel; 2). Chirp-like signal issued in multi-channels; 3). Step-like and random signals issued for validation. Example of chirp-like signal Chirp-like signal and corresponding responses 12

13 Modeling Test Flights Flight testing for modeling purpose 13

14 Modeling Model Structure Hover Model of HeLion 14

15 Modeling Parameter Identification Parameter Identification: 1). Angular rate dynamics; 2). Horizontal velocity dynamics; 3). Yaw dynamics; 4). Heave dynamics. 15

16 Model Validation Validation for the identified Model: Eigenvalues and corresponding mode description 16

17 Model Validation 17

18 Structure of Automatic Flight Control System ( x, y, z, ψ ) ( u, v, w, r) δ, δ, δ, δ ) ( a e u r helicopter state Flight Schedule Outer-loop Control Inner-loop Control 18

19 Automatic Inner Loop Control Using CNF Control Technique The newly developed CNF (composite nonlinear feedback) control technique is capable of tracking a target reference with fast settling time and with small or zero overshoot. It has many advantages compared to conventional approaches 1.5 Response with a small damping ratio 1 fast rise time Magnitude Response with a large damping ratio no overshoot 0.5 CNF Control Time 19

20 CNF Control Law 20

21 Comparison with Conventional LQR Technique Results were obtained from actual flight tests 21

22 Full Envelop Flight Paths Illustration 1. standby 2. takeoff 7. wheeling 8. backward down spiraling 3. hovering 4. slithering 5. head turning 6. pirouette 9. hovering 10. landing 11. standby 20 m 15 m 20 m 15 m 12 m 60 m 12 m A 30 m 15 m B 15 m 15 m yh O y A x x 22

23 Video Demo of Full Envelop Flight 23

24 Target Aiming Test Illustration 24

25 Video Demo of Target Aiming 25

26 Current Work Ground Target Tracking & Attacking 26

27 Experimental Results of Ground Target Detection Single stage pattern recognition Dual stage pattern recognition + : predicted position : measured position 27

28 Summary of Results for Ground Target Detection Number of images Number of errors Accuracy Single stage pattern recognition % Flying training data Dual Stage pattern recognition % Single stage pattern recognition % Flying testing data Dual stage pattern recognition % 28

29 Future Work Formation Flight 29

30 Acknowledgement Special thanks to Defence Science & Technology Agency (DSTA) National University of Singapore (NUS) Temasek Laboratories And My Research Team at NUS Professor T.H. Lee, Dr Lum Kai Yew, Dr Peng Kemao Dr Dong Miaobo, Dr Lan Weiyao Mr. Cai Guowei, Mr. Lin Feng & Mr. Yun Ben 30

31 Acknowledgement As well as in memory of our first chopper, NUSIX, which crashed twice on 25 March 2006 and 15 February

32 Thank you! 32