Learning Diverse Skills for Behavior Models with Mixture of Experts

TL;DR

This paper introduces Di-BM, a mixture of experts approach for imitation learning that enables behavior models to learn diverse, specialized skills, improving multi-task performance and data efficiency in robotic manipulation.

Contribution

The paper proposes Di-BM, a novel mixture of experts framework with energy-based models for multi-task imitation learning, enhancing specialization and transferability of skills.

Findings

Di-BM outperforms state-of-the-art baselines on robotic tasks.

Fine-tuning Di-BM on new tasks is more data-efficient.

Experts specialize in different observation sub-regions, reducing interference.

Abstract

Imitation learning has demonstrated strong performance in robotic manipulation by learning from large-scale human demonstrations. While existing models excel at single-task learning, it is observed in practical applications that their performance degrades in the multi-task setting, where interference across tasks leads to an averaging effect. To address this issue, we propose to learn diverse skills for behavior models with Mixture of Experts, referred to as Di-BM. Di-BM associates each expert with a distinct observation distribution, enabling experts to specialize in sub-regions of the observation space. Specifically, we employ energy-based models to represent expert-specific observation distributions and jointly train them alongside the corresponding action models. Our approach is plug-and-play and can be seamlessly integrated into standard imitation learning methods. Extensive experiments on multiple real-world robotic manipulation tasks demonstrate that Di-BM significantly outperforms state-of-the-art baselines. Moreover, fine-tuning the pretrained Di-BM on novel tasks exhibits superior data efficiency and the reusable of expert-learned knowledge. Code is available at https://github.com/robotnav-bot/Di-BM.