Approximate Model-Based Shielding for Safe Reinforcement Learning
This addresses safety-critical applications of RL, offering a method to mitigate worst-case risks, though it is incremental as it builds on existing shielding approaches.
The paper tackles the problem of ensuring safety in reinforcement learning (RL) for real-world systems by proposing approximate model-based shielding (AMBS), a look-ahead algorithm that verifies policies against safety constraints without prior knowledge of safety dynamics, and demonstrates superior performance on Atari games with safety labels.
Reinforcement learning (RL) has shown great potential for solving complex tasks in a variety of domains. However, applying RL to safety-critical systems in the real-world is not easy as many algorithms are sample-inefficient and maximising the standard RL objective comes with no guarantees on worst-case performance. In this paper we propose approximate model-based shielding (AMBS), a principled look-ahead shielding algorithm for verifying the performance of learned RL policies w.r.t. a set of given safety constraints. Our algorithm differs from other shielding approaches in that it does not require prior knowledge of the safety-relevant dynamics of the system. We provide a strong theoretical justification for AMBS and demonstrate superior performance to other safety-aware approaches on a set of Atari games with state-dependent safety-labels.