Dual RL: Unification and New Methods for Reinforcement and Imitation Learning

TL;DR

This paper introduces a unified dual RL framework that simplifies optimization and proposes new methods ReCOIL and f-DVL to improve offline imitation learning and reinforcement learning, validated on robotic tasks.

Contribution

It unifies existing offline RL and IL algorithms under dual RL, identifies their limitations, and proposes novel methods to overcome these issues.

Findings

ReCOIL achieves near-expert imitation from arbitrary data.

f-DVL improves training stability over XQL in offline RL.

Both methods outperform prior approaches on robotic tasks.

Abstract

The goal of reinforcement learning (RL) is to find a policy that maximizes the expected cumulative return. It has been shown that this objective can be represented as an optimization problem of state-action visitation distribution under linear constraints. The dual problem of this formulation, which we refer to as dual RL, is unconstrained and easier to optimize. In this work, we first cast several state-of-the-art offline RL and offline imitation learning (IL) algorithms as instances of dual RL approaches with shared structures. Such unification allows us to identify the root cause of the shortcomings of prior methods. For offline IL, our analysis shows that prior methods are based on a restrictive coverage assumption that greatly limits their performance in practice. To fix this limitation, we propose a new discriminator-free method ReCOIL that learns to imitate from arbitrary off-policy data to obtain near-expert performance. For offline RL, our analysis frames a recent offline RL method XQL in the dual framework, and we further propose a new method f-DVL that provides alternative choices to the Gumbel regression loss that fixes the known training instability issue of XQL. The performance improvements by both of our proposed methods, ReCOIL and f-DVL, in IL and RL are validated on an extensive suite of simulated robot locomotion and manipulation tasks. Project code and details can be found at this https://hari-sikchi.github.io/dual-rl.