Skip to content

Control Overview

Simply put, pure_pursuit.py and pure_pursuit.cpp are two ROS (Robot Operating Software) files which make up the two implementations of Formula Trinity's AI Control.

These ROS nodes listen for the car's pose (orientation and location of the car: car_x, car_y, quaternion → car_yaw) and a stream of race-line waypoints, pick a goal point a short distance in front of the car (the lookahead circle), compute a steering angle to reach that goal, and publish an Ackermann drive command (steer + accel).

Subscribed Topics

  • /pose — where the car is and how it is oriented.
    Used to set car_x, car_y, car_yaw.

  • /raceline (PointArray) — the sequence of (x, y) waypoints to follow.

  • Speed Source — either
    /odometry_integration/car_state for sim velocity, or
    /ros_can/wheel_speeds for rear-wheel-based speed velocity.

  • Control overrides from the state machine —
    /state_machine/safety_control and /state_machine/mission_control
    (stop/crawl/target velocity/accel).

Published Topics

  • /cmd (AckermannDriveStamped) — the actual steering angle and acceleration to the vehicle.
  • /steering_angle (Pose) for visualization and /distance (how far you've travelled).
    The C++ version also publishes a marker for the goal point.

Important Parameters

  • use_slam (bool) — read pose from SLAM frames vs base_footprint; sets frame_id.
  • nominal_speed, crawl_speed (m/s) — baseline speeds.
  • linear_speed_source — "sim" or "wheel_speeds".
  • wheel_radius (m) — used to convert wheel rpm to m/s.
  • Lookahead distance ld — computed as L * coefficient (wheelbase × multiplier).
    In pure_pursuit.py it's a fixed self.L * 1.2.
    In pure_pursuit.cpp it's parameterized via ld_coefficient.
    This is the #1 fine-tuning knob.

Glossary

  • Ackermann — car-like steering (front-wheels turn), so we command steering angle and acceleration.
  • Lookahead (ld) — radius of circle centered at the car; we aim for a point where the raceline crosses the circle. Bigger ld looks further ahead.
  • Alpha (α) — the heading error between where the car points and where the goal point is.
  • LAP / beta / gamma (C++) — small add-ons to pure pursuit to account for curvature and how far off the path you are; they slightly shift the goal point for smoother tracking.

Resources

Here's some helpful videos that describe what we typically deal with in Control.

Feedback Control: https://www.youtube.com/watch?v=O-OqgFE9SD4&t=21s&pp=ygUQZmVlZGJhY2sgY29udHJvbA%3D%3D

PID Introduction: https://www.youtube.com/watch?v=UR0hOmjaHp0&pp=ygUQZmVlZGJhY2sgY29udHJvbA%3D%3D

PID Demo: https://youtu.be/fusr9eTceEo

PID Steering: https://youtu.be/4Y7zG48uHRo

Note, we don't use PID's for steering, but its a good illustration of how they work

Pure Pursuit (long video) https://youtu.be/x9s8J4ucgO0

For more info on C++ code please see: - Migration to C++