Learning Safety Constraints from Demonstrations with Unknown Rewards
This addresses the challenge of ensuring safety in reinforcement learning for domains like autonomous driving, where demonstrations may have different rewards, and is incremental by extending prior work limited to known rewards or dynamics.
The paper tackles the problem of learning safety constraints from demonstrations with unknown rewards and unknown environment dynamics, proposing CoCoRL which constructs a convex safe set that provably guarantees safety and converges to the true safe set for near-optimal demonstrations, leading to safe driving behavior in evaluations.
We propose Convex Constraint Learning for Reinforcement Learning (CoCoRL), a novel approach for inferring shared constraints in a Constrained Markov Decision Process (CMDP) from a set of safe demonstrations with possibly different reward functions. While previous work is limited to demonstrations with known rewards or fully known environment dynamics, CoCoRL can learn constraints from demonstrations with different unknown rewards without knowledge of the environment dynamics. CoCoRL constructs a convex safe set based on demonstrations, which provably guarantees safety even for potentially sub-optimal (but safe) demonstrations. For near-optimal demonstrations, CoCoRL converges to the true safe set with no policy regret. We evaluate CoCoRL in gridworld environments and a driving simulation with multiple constraints. CoCoRL learns constraints that lead to safe driving behavior. Importantly, we can safely transfer the learned constraints to different tasks and environments. In contrast, alternative methods based on Inverse Reinforcement Learning (IRL) often exhibit poor performance and learn unsafe policies.