Accelerating Safe Reinforcement Learning with Constraint-mismatched Policies
arXiv:2006.11645
Abstract
We consider the problem of reinforcement learning when provided with (1) a baseline control policy and (2) a set of constraints that the learner must satisfy. The baseline policy can arise from demonstration data or a teacher agent and may provide useful cues for learning, but it might also be sub-optimal for the task at hand, and is not guaranteed to satisfy the specified constraints, which might encode safety, fairness or other application-specific requirements. In order to safely learn from baseline policies, we propose an iterative policy optimization algorithm that alternates between maximizing expected return on the task, minimizing distance to the baseline policy, and projecting the policy onto the constraint-satisfying set. We analyze our algorithm theoretically and provide a finite-time convergence guarantee. In our experiments on five different control tasks, our algorithm consistently outperforms several state-of-the-art baselines, achieving 10 times fewer constraint violations and 40% higher reward on average.
arXiv commentInternational Conference on Machine Learning (ICML) 2021
References in corpus (14)
- Continuous control with deep reinforcement learning
- Playing Atari with Deep Reinforcement Learning
- Trust Region Policy Optimization
- Leveraging Demonstrations for Deep Reinforcement Learning on Robotics Problems with Sparse Rewards
- Generative Adversarial Imitation Learning
- Deep Q-learning from Demonstrations
- Safe Exploration in Continuous Action Spaces
- Reward Constrained Policy Optimization
- Benchmarking Batch Deep Reinforcement Learning Algorithms
- Lyapunov-based Safe Policy Optimization for Continuous Control
- Reinforcement Learning from Imperfect Demonstrations
- Learning to be Safe: Deep RL with a Safety Critic
- Agent-Agnostic Human-in-the-Loop Reinforcement Learning
- On-Policy Robot Imitation Learning from a Converging Supervisor