Table of contents

Towards Bridging the Sim-to-Real Gap - F1TENTH Gym Environment for Learning-based Control Policies in Autonomous Racing

Speaker: Kukjin Jang

Abstract: The F1TENTH autonomous racing platform provides an effective approach for developing perception, planning, and control algorithms for autonomous driving. The F1TENTH-GYM simulation environment further enables the rapid experimentation of learning-based control policies. This talk presents an overview of the environment and a case study using reinforcement learning and imitation learning to learn control policies for the F1TENTH race car. Experimental results of applying the policies both in simulation and on scaled real-world experiments are presented as well as future research directions for bridging the sim-to-real gap.