Extrapolating Beyond Suboptimal Demonstrations via Inverse Reinforcement Learning from Observations

TL;DR

T-REX is a novel reward-learning algorithm that extrapolates beyond suboptimal demonstrations to infer high-quality reward functions, enabling reinforcement learning agents to outperform the demonstrators in benchmark tasks.

Contribution

Introduces T-REX, a reward extrapolation method that leverages ranked demonstrations to infer better reward functions than the demonstrator's performance.

Findings

T-REX outperforms state-of-the-art imitation learning and IRL methods.

T-REX achieves more than twice the performance of the best demonstration.

T-REX is robust to ranking noise and can infer intentions from noisy improvement observations.

Abstract

A critical flaw of existing inverse reinforcement learning (IRL) methods is their inability to significantly outperform the demonstrator. This is because IRL typically seeks a reward function that makes the demonstrator appear near-optimal, rather than inferring the underlying intentions of the demonstrator that may have been poorly executed in practice. In this paper, we introduce a novel reward-learning-from-observation algorithm, Trajectory-ranked Reward EXtrapolation (T-REX), that extrapolates beyond a set of (approximately) ranked demonstrations in order to infer high-quality reward functions from a set of potentially poor demonstrations. When combined with deep reinforcement learning, T-REX outperforms state-of-the-art imitation learning and IRL methods on multiple Atari and MuJoCo benchmark tasks and achieves performance that is often more than twice the performance of the best demonstration. We also demonstrate that T-REX is robust to ranking noise and can accurately extrapolate intention by simply watching a learner noisily improve at a task over time.