Establishing and Maintaining Formations of Mini Quadrotors Audrow J. Nash, Cory M. Engel, James M. Conrad William Lee College of Engineering University of North Carolina at Charlotte Charlotte, North Carolina
Abstract: What? Quadrotor system design – On-board processing Flight algorithm for motion planning – Can be expanded for swarms Start of solo quadrotor autonomous flight
Abstract: Why? Develop a base for swarm application, which has the following benefits: – Payload manipulation – Surveillance application – Search-and-Rescue
Abstract: Novel Contribution On-board processing (no external localization system) Use of Wii Camera for localization and motion planning
Table of contents 1.System Overview 2.Motion Planning Algorithm 3.Characterizing Vision System 4.Vision System Testing 5.Autonomous Flight Progress 6.Conclusion and Future Work 7.Contact Information
Overview: System Flow Chart
Overview: Wii Camera Monochrome 1024x768 camera – After initial resolution of 128x96 undergoes 8x subpixel analysis Tracks the four highest intensity light sources Outputs light source Cartesian coordinate (X, Y) and observed size Combined with infrared (IR) pass filter (extracted from Wiimote)
Overview: System Flow Chart
Overview: Reference Beacon Composed of 4 IR LEDs B1, B2, B3 for localization * Discussed later B4 to saturate camera capabilities – Reduce environmental noise
Overview: System Flow Chart
Overview: Vision Processor Receives Wii camera Data – Uses I2C Outputs movement command to flight system – PWM to mimic remote transmitter
Overview: System Flow Chart
Overview: Flight System Interprets movement commands from vision system and actuates accordingly Handles stable flight with input from sensors
Overview: System Flow Chart
System Overview: Sensors SensorPurpose Inertial Measurement Unit (IMU) Level/stable flight Sonar sensorMeasure altitude Wii Camera is not mentioned because it does not pass information to the flight system directly *
Overview: System Flow Chart
System Overview: Actuators
Motion Planning Algorithm: Overview
Motion Planning Algorithm: Code //Main loop void loop() { GetAverageBlobXYCoordinates(); if (IsLogical()) SendMovementCommand(); } Flight system will hover quadrotor until given a movement command to actuate *
Motion Planning Algorithm: NonOrthagonal Case Begins by flying to correct altitude – Beacon is half of possible Y value Alternates between adjusting yaw and roll (lateral movement) – Allows for constant site of beacon
Motion Planning Algorithm: Orthagonal Case Behavior begins when dobser:B1;B2 == dobser:B2;B3 Distance of the quadrotor from the beacon is proportional to the perceived distance between the first (B1)and third (B3)light source
Characterizing Vision System: Data Quadrotor distance from beacon Distance between outer light sources
Characterizing Vision System: Results Used equations to develop straight line approximations to determine quadrotor distance from reference beacon
Characterizing Vision System: Testing It was observed that the algorithm can direct a body to a position reliably with respect to the beacon for roll, pitch, and yaw
Vision System Testing: Beacon Configuration Results: Greater than 4 light sources is unreliable Additional check of beacon is necessary Mitigation: Increased filtering Directional nature of LEDs
Vision System Testing: Environmental Lighting Impact Results: Indoor lighting conditions have no noticeable effect on vision system Additional testing for outdoor conditions recommended
Autonomous Flight Progress Wireless call flight modes – Safe Mode: toggle to arm and disarm quadrotor – Stabilize: quadrotor flight remains level – Altitude hold: hover a specific height Vision system mimics transmitter signal to Flight system Role and pitch response to beacon
Conclusion Accomplished: – Quadrotor system – Algorithm for autonomous flight of a single quadrotor with on-board processing – Tested algorithm – Made progress towards autonomous flight
Future work Reliable autonomous liftoff/landing Increased logical checks of beacon Develop and test a swarm algorithm
Contact Info Audrow J. Nash, Cory M. Engel, James M. Conrad William Lee College of Engineering University of North Carolina at Charlotte Charlotte, North Carolina
Processor Specifications Vision SystemFlight System BoardRed BoardAll In One Pro V2.0 ProcessorATMega 328ATMega 2560 Flash32kB256kB Clock speed16MHz GPIO/Analog Inputs14/686/16 Other6 axis gyro/accel