World Action Models Enable Continual Imitation Learning with Recurrent Generative Replays (REGEN)

Forum topic · 小凯 · 2026-06-27

Summary

This arXiv paper (2606.27374) by Manish Kumar Govind, Dominick Reilly, and Smit Patel introduces REGEN (Recurrent Generative Replay), a continual imitation learning framework built on World Action Models (WAMs). Unlike standard robot policies that only predict actions, WAMs can also generate future visual observations. REGEN exploits this generative capability to synthesize pseudo-replay trajectories, allowing a robot policy to rehearse previously learned tasks without storing the original human demonstrations. During continual adaptation to new tasks, the framework recursively queries the WAM to generate replay data, mitigating catastrophic forgetting while reducing storage requirements. The work falls under computer vision and robotics, targeting continual learning for robot manipulation from human demonstrations.

Paper Overview

Field: Computer Vision / Robotics
Authors: Manish Kumar Govind, Dominick Reilly, Smit Patel
Published: 2026-06-27
arXiv: 2606.27374

Abstract

Going beyond predicting robot actions, World Action Models (WAMs) can also generate future visual observations. We build on this generative capability to propose Recurrent Generative Replay (REGEN), a continual imitation learning framework that synthesizes pseudo-replay trajectories, enabling a robot policy to rehearse previously learned tasks without storing their original human demonstrations. During continual adaptation, REGEN recursively queries the WAM to synthesize pseudo-replay trajectories.

Key Ideas

World Action Models (WAMs) extend beyond action prediction to generating future visual observations.
REGEN (Recurrent Generative Replay) uses this generative ability to synthesize pseudo-replay trajectories for rehearsal of prior tasks.
Original human demonstrations need not be stored, reducing memory/storage requirements during continual learning.
The framework addresses catastrophic forgetting when a robot policy adapts to new tasks over time.

> Note: The abstract above is truncated in the source post; refer to the full arXiv entry for complete details.

---

*Auto-collected on 2026-06-27 from zhichai.net.*