Model-Based Imitation Learning Using Entropy Regularization of Model and Policy

TL;DR

This paper introduces MB-ERIL, a model-based imitation learning method with entropy regularization that enhances sample efficiency by using two discriminators to distinguish between expert and generated actions and transitions, demonstrated through simulations and robot experiments.

Contribution

The paper proposes a novel entropy-regularized model-based imitation learning framework with dual discriminators for improved sample efficiency and learning effectiveness.

Findings

MB-ERIL achieves competitive performance in imitation tasks.

Significantly improves sample efficiency over baseline methods.

Effective in both simulations and real robot experiments.

Abstract

Approaches based on generative adversarial networks for imitation learning are promising because they are sample efficient in terms of expert demonstrations. However, training a generator requires many interactions with the actual environment because model-free reinforcement learning is adopted to update a policy. To improve the sample efficiency using model-based reinforcement learning, we propose model-based Entropy-Regularized Imitation Learning (MB-ERIL) under the entropy-regularized Markov decision process to reduce the number of interactions with the actual environment. MB-ERIL uses two discriminators. A policy discriminator distinguishes the actions generated by a robot from expert ones, and a model discriminator distinguishes the counterfactual state transitions generated by the model from the actual ones. We derive structured discriminators so that the learning of the policy and the model is efficient. Computer simulations and real robot experiments show that MB-ERIL achieves a competitive performance and significantly improves the sample efficiency compared to baseline methods.