Dvij Kalaria, Shreya Sharma, Sarthak Bhagat, Haoru Xue, John Dolan
Robotics Institute, Carnegie Mellon University
Abstract
Off-road navigation is a challenging problem both at the planning level to get a smooth trajectory and at the control level to avoid flipping over, hitting obstacles or getting stuck at a rough patch. There have been several recent works on using classical approaches involving depth map prediction following by smooth trajectory planning and using controller to track it. We design an end-to-end reinforcement learning (RL) system for an autonomous vehicle in off-road environments using a custom designed simulator in Unity game engine. We warm start the agent by imitating a rule-based controller and use Proximal Policy Optimization (PPO) to improve the policy based on a reward that incorporates Control Barrier Functions (CBF), facilitating the agent's ability to generalize effectively to real-world scenarios. The training involves agents concurrently undergoing domain-randomized trials in various environments. We also deploy our trained model on a real buggy RC car.
Simulation results
RC car results
BibTeX
@misc{wroom2023,
title={WROOM: An Autonomous Driving Approach for Off-Road Navigation},
author={Dvij Kalaria, Sarthak Bhagat, Shreya Sharma, Haoru Xue},
year={2023},
}