Reinforcement Learning from Imperfect Demonstrations

TL;DR

The paper introduces NAC, a unified reinforcement learning algorithm that effectively leverages imperfect demonstrations and environment interactions to learn robust policies, outperforming existing methods in driving simulations.

Contribution

NAC is a novel unified algorithm that normalizes Q-values, enabling robust learning from imperfect demonstrations without combining separate supervised and reinforcement losses.

Findings

NAC outperforms existing baselines in driving game simulations.

NAC is robust to noisy and suboptimal demonstration data.

NAC surpasses demonstrator performance through interactive refinement.

Abstract

Robust real-world learning should benefit from both demonstrations and interactions with the environment. Current approaches to learning from demonstration and reward perform supervised learning on expert demonstration data and use reinforcement learning to further improve performance based on the reward received from the environment. These tasks have divergent losses which are difficult to jointly optimize and such methods can be very sensitive to noisy demonstrations. We propose a unified reinforcement learning algorithm, Normalized Actor-Critic (NAC), that effectively normalizes the Q-function, reducing the Q-values of actions unseen in the demonstration data. NAC learns an initial policy network from demonstrations and refines the policy in the environment, surpassing the demonstrator's performance. Crucially, both learning from demonstration and interactive refinement use the same objective, unlike prior approaches that combine distinct supervised and reinforcement losses. This makes NAC robust to suboptimal demonstration data since the method is not forced to mimic all of the examples in the dataset. We show that our unified reinforcement learning algorithm can learn robustly and outperform existing baselines when evaluated on several realistic driving games.