mahaitongdae/safe_exp_env

Thrid-party safety-gym safe exploration environments for autonomous driving. Also environment used for IROS 2021 paper: Model-based Constrained Reinforcement Learning using Generalized Control Barrier Function

Safe Exploration Env

Code for paper Model-based Constrained Reinforcement Learning using Generalized Control Barrier Function submitted to 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2021). An OpenAI safety-gym third-party environment with SUMO.

This repo is modified from the original version.

System Requirements

• Ubuntu 16.04 or 18.04
• SUMO <= 1.7.0
• safety-gym environment

Related Repos and Papers

• Integrated Decision and Control Scheme for Autonomous Driving and Corresponding training environments.

This paper provides a integrated decision and control scheme for autonomous vehicles, and provides the major supporting techniques for the real vehicle experiment in this paper.

@article{guan2021integrated,
  title={Integrated Decision and Control: Towards Interpretable and Efficient Driving Intelligence},
  author={Guan, Yang and Ren, Yangang and Li, Shengbo Eben and Ma, Haitong and Duan, Jingliang and Cheng, Bo},
  journal={arXiv preprint arXiv:2103.10290},
  year={2021}
}

• OpenAI safety starter agents