In this lecture, the focus is on designing a Pure Pursuit controller for autonomous race cars to track a reference path through the environment. Pure Pursuit is a geometric path tracking controller that relies on the kinematic vehicle model to select steering commands. It is particularly suited for lateral control, ignoring dynamic forces on the vehicle.

Key points from the lecture:

Introduction to Pure Pursuit: Pure Pursuit is a geometric path tracking controller that utilizes the geometry of the vehicle's kinematics and a reference path. It is a lateral controller that ignores dynamic forces and assumes a no-slip condition at the wheels. Assumption:

• Vehicle is given a sequence of 2D positions ( waypoints to follow)

• Vehicle knows where the given waypoints are in the vehicle’s frame of reference

  • Underlying assumptions that vehicle can localize itself

• Goal is to follow these waypoints

  

Algorithm Overview: The goal is to enable autonomous race cars to track a specific race line or path quickly. The lecture discusses how to use the Pure Pursuit algorithm to achieve faster racing times compared to reactive algorithms.

Vehicle Planning and Control Stack: The lecture covers the overall architecture of the autonomous vehicle planning and control stack, including perception, planning, and actuation modules. Different planners, including mission, behavioral, and local planners, work together to generate paths and trajectories.

Pure Pursuit Algorithm Steps: The algorithm starts with a sequence of waypoints in a global coordinate frame. It calculates a reference point (target point) at a fixed look-ahead distance on the path. The algorithm then computes the curvature of the arc that connects the vehicle to the target point. The steering angle is adjusted based on the curvature to make the vehicle follow the desired path.

ICR - Instantaneous Center of Rotation

Goal Point Selection: Different strategies are discussed for selecting the goal point when waypoints are not exactly on the path. Closest point inside the circle, closest point outside the circle, and interpolation between adjacent waypoints are explored.

Tuning and Performance: Parameters, particularly the look-ahead distance, play a significant role in tuning the Pure Pursuit algorithm. Smaller look-ahead distances lead to more aggressive maneuvers, while larger distances result in smoother trajectories. Velocity components at waypoints can be hand-tuned to adapt behavior based on speed.

Implementation Steps: The lecture provides practical implementation steps for using the Pure Pursuit algorithm in an autonomous vehicle. Creating a map, defining waypoints, smoothing paths, localizing using particle filter, and updating waypoints are discussed.

Limitations and Considerations: The algorithm assumes no-slip conditions and does not account for dynamic forces. For aggressive maneuvers, understanding tire limits and considering more advanced control strategies is necessary.

In summary, the Pure Pursuit lecture focuses on the design and implementation of a geometric path tracking controller for autonomous race cars. It covers algorithm basics, goa