IED MiniProject: Autonomous Control

1
IED Mini-ProjectAutonomous Control

• David Brigada

2
Overview

• Types of control considered
• Approaches to autonomous control
• Overview of selected system
• Implementation for mini-project
• Testing results
• Demonstration

3
Types of Control Considered

• Tethered remote control
• Human interactivity
• Tether issues
• Radio control
• Human interactivity
• More complex
• Radio interference

4
Approaches toAutonomous Control

• Vision
• Interactive dynamic response
• Computationally intensive
• Difficult to get right
• Sound
• Interactive dynamic response
• Small amounts of data
• Easily get lost

5
More Approaches toAutonomous Control

• GPS
• Non-interactive
• Provides position information
• Not reliable inside buildings
• Limited precision

6
Overview of Selected System

• Sound Trilateration
• Similar to GPS
• Beacons deployed around corners of maze
• Software decodes position from relative signal
  timings

7
Sound Trilateration
8
Use of Position Information

• Algorithm uses timings to find
• Position
• Bearing
• Speed
• Algorithm is programmed with a set of points
• Waypoints lead a path through maze
• Algorithm follows waypoints one after another
• Actions can be programmed at waypoints
• Collect or deposit balls

9
Limitations of Selected System

• No response to dynamic environment
• Handling balls that have moved
• Interacting with other robot
• Not a common solution to this problem
• Possible approaches
• Add sensors to detect environment change
• Add manual override radio control

10
Implementation forMini-Project

• Interface access limitations time constraint
• Difficulty of testing
• Solution
• Simulate the control algorithm and environment
• Client/server model
• Server Simulates physics of maze environment
• Also displays results for interpretation
• Client Runs control algorithm

11
Client program

• Design like intended target system
• Contains all the functionality of the control
  algorithm
• Takes differential time inputs
• Produces steering and drive outputs

12
Server program

• Simulates the environment
• Calculates timing differences for dispatch to
  control algorithm
• Calculates kinematics of robot motion
• Displays output
• 3 dimensional output rendered in OpenGL

13
Client and ServerCommunications

• Client sends drive motor duty cycle and servo
  motor pulsewidth
• Server sends beacon timing differences
• Server does NOT send position, bearing, or speed
  directly to client

Signal timings
Server
Client
Output waveforms
14
Testing Results

• Tests were done to ensure accuracy of robot
  response
• Control algorithm was continually able to
  reproduce its movements within very small
  fractions of an inch
• Control algorithm maintained integrity to ½ inch
  until system precision was reduced to one
  twentieth of nominal value
• Unknown noise levels from inputs

15
Demonstration

• The following is a presentation of the simulation
  in action.
• The system picks random points to visit for this
  simulation.