M2Distill: Multi-Modal Distillation for Lifelong Imitation Learning

TL;DR

M2Distill introduces a multi-modal distillation approach that maintains a consistent latent space across vision, language, and actions, enabling effective lifelong imitation learning without catastrophic forgetting.

Contribution

It presents a novel multi-modal distillation method that preserves latent space consistency across modalities during incremental learning, improving lifelong imitation performance.

Findings

Outperforms prior methods on LIBERO benchmarks

Maintains task performance across incremental learning steps

Effectively integrates new skills without forgetting

Abstract

Lifelong imitation learning for manipulation tasks poses significant challenges due to distribution shifts that occur in incremental learning steps. Existing methods often focus on unsupervised skill discovery to construct an ever-growing skill library or distillation from multiple policies, which can lead to scalability issues as diverse manipulation tasks are continually introduced and may fail to ensure a consistent latent space throughout the learning process, leading to catastrophic forgetting of previously learned skills. In this paper, we introduce M2Distill, a multi-modal distillation-based method for lifelong imitation learning focusing on preserving consistent latent space across vision, language, and action distributions throughout the learning process. By regulating the shifts in latent representations across different modalities from previous to current steps, and reducing discrepancies in Gaussian Mixture Model (GMM) policies between consecutive learning steps, we ensure that the learned policy retains its ability to perform previously learned tasks while seamlessly integrating new skills. Extensive evaluations on the LIBERO lifelong imitation learning benchmark suites, including LIBERO-OBJECT, LIBERO-GOAL, and LIBERO-SPATIAL, demonstrate that our method consistently outperforms prior state-of-the-art methods across all evaluated metrics.