English

Safe Model-Based Reinforcement Learning with an Uncertainty-Aware Reachability Certificate

Robotics 2022-10-17 v1 Machine Learning Systems and Control Systems and Control

Abstract

Safe reinforcement learning (RL) that solves constraint-satisfactory policies provides a promising way to the broader safety-critical applications of RL in real-world problems such as robotics. Among all safe RL approaches, model-based methods reduce training time violations further due to their high sample efficiency. However, lacking safety robustness against the model uncertainties remains an issue in safe model-based RL, especially in training time safety. In this paper, we propose a distributional reachability certificate (DRC) and its Bellman equation to address model uncertainties and characterize robust persistently safe states. Furthermore, we build a safe RL framework to resolve constraints required by the DRC and its corresponding shield policy. We also devise a line search method to maintain safety and reach higher returns simultaneously while leveraging the shield policy. Comprehensive experiments on classical benchmarks such as constrained tracking and navigation indicate that the proposed algorithm achieves comparable returns with much fewer constraint violations during training.

Keywords

Cite

@article{arxiv.2210.07553,
  title  = {Safe Model-Based Reinforcement Learning with an Uncertainty-Aware Reachability Certificate},
  author = {Dongjie Yu and Wenjun Zou and Yujie Yang and Haitong Ma and Shengbo Eben Li and Jingliang Duan and Jianyu Chen},
  journal= {arXiv preprint arXiv:2210.07553},
  year   = {2022}
}

Comments

12 pages, 6 figures

R2 v1 2026-06-28T03:37:19.795Z