watonomous.github.io

1. Software Division
2. Software Division Home
3. Path Planning Group

[ Software Division : Feedback Controllers ]

Created by [ Rowan Dempster] on Dec 27, 2019

High Level Objectives

Even if we have an ideal path for the car to follow, actually getting a mechanical robot to follow that path is not trivial, since there are always random errors that accumulate, like if the car ran over a small pebble and deviated slightly. To solve this, feedback controllers have to be made that run at an even higher frequency than the local planner, attempting to steer the car towards its target path despite all the tiny errors constantly accumulating.

Implementation

Inputs

State of the world and the vehicle, as well as the target path generated by the Trajectory Planner

Outputs

Where to turn the wheels and how fast to move them

ROS Topics/Msgs Integration

Inputs

• VehicleState.msg comprising of
  • x coordinate
  • y coordinate
  • theta (car's orientation)
  • forward_speed
• Path.msg is an array of VehicleStates

Output

• DesiredOutput.msg
  • theta
  • torque
  • state (stopped/driving)

Year 1 Implementation

Speed Controller

See https://git.uwaterloo.ca/WATonomous/decision/blob/feedback/path_planning/feedback/speed_controller.cpp

Wheel Angle Controller

See https://git.uwaterloo.ca/WATonomous/decision/blob/feedback/path_planning/feedback/wheel_angle_controller.cpp

Learning Resources

• Relevant webinar (look for Feedback Controllers): https://drive.google.com/drive/folders/1AZmyoGmbJTf1ssVKEielogkGBdnxP_pJ
• PID Control https://www.youtube.com/watch?v=UR0hOmjaHp0

Team members

Abhinav Grover (Lead)
Edwin Lo
Syed Moiz
Spencer Zhou

Document generated by Confluence on Dec 10, 2021 04:01

Atlassian