Learning the Subsystem of Local Planning for Autonomous Racing

TL;DR

This paper presents a hierarchical planning architecture that combines a high level planner and path following system with a reinforcement learning agent that learns that subsystem of obstacle avoidance and can achieve faster average times compared to the baseline end-to-end planner and a 94% success rate which is similar toThe baseline.

Abstract

The problem of autonomous racing is to navigate through a race course as quickly as possible while not colliding with any obstacles. We approach the autonomous racing problem with the added constraint of not maintaining an updated obstacle map of the environment. Several current approaches to this problem use end-to-end learning systems where an agent replaces the entire navigation pipeline. This paper presents a hierarchical planning architecture that combines a high level planner and path following system with a reinforcement learning agent that learns that subsystem of obstacle avoidance. The novel "modification planner" uses the path follower to track the global plan and the deep reinforcement learning agent to modify the references generated by the path follower to avoid obstacles. Importantly, our architecture does not require an updated obstacle map and only 10 laser range finders to avoid obstacles. The modification planner is evaluated in the context of F1/10th autonomous racing and compared to a end-to-end learning baseline, the Follow the Gap Method and an optimisation based planner. The results show that the modification planner can achieve faster average times compared to the baseline end-to-end planner and a 94% success rate which is similar to the baseline.

Figures (11)

• Figure 1: Illustration of Hybrid Planning Approach: The green dotted line is the global plan, the blue blocks are the obstacles and the red line is the path taken by the vehicle. A path following algorithm that can follow a path is combined with an RL agent that learns the subsystem of obstacle avoidance to create the modification planner.
• Figure 2: Proposed Local Planner Architecture: A path follower to track the reference path is combined in parallel with an RL agent that modifies the path follower references.
• Figure 3: Planning, Perception, and Control Navigation Stack
• Figure 4: Pure Pursuit Path Follower:$\alpha$ is the angle to the goal, $l_d$ is the look ahead distance, $L$ is the vehicle wheelbase, and $R$ is the turning radius.
• Figure 5: Reinforcement Learning Agent Interacting With Environment