Diffusion ReRoll: Revisable Denoising for Robotic Sequential Prediction

Abstract

We propose Diffusion ReRoll, a diffusion-based framework for robotic sequential prediction that enables revisable denoising over horizons. Existing diffusion-based sequence predictors typically perform a single monotonic denoising process. In contrast, Diffusion ReRoll selectively re-noises regions that reach a predefined noise threshold while the remaining regions continue denoising, so the re-noised regions can be refined again using context from the rest of the horizon. This structured re-noising enables iterative cross-horizon revision, allowing earlier and later segments to revise one another, while maintaining local consistency. We evaluate Diffusion ReRoll against full-sequence diffusion and causal denoising based on Diffusion Forcing across long-horizon planning, policy learning, unified video-action modeling, and real-world navigation. On OGBench PointMaze and AntMaze, Diffusion ReRoll achieves relative gains in average success rate of 21% over Diffusion Forcing in matched guidance-based planning and 23% over Diffuser in goal-inpainting. In diffusion-policy-style action prediction, Diffusion ReRoll improves average success by 56.5% relative to Diffusion Policy across different prediction horizons and history lengths on LIBERO-10 under the official reset protocol. In unified video-action prediction, Diffusion ReRoll improves policy and inverse dynamics performance, especially under out-of-distribution evaluation, and achieves the best action-video consistency. Real-world navigation further demonstrates improved success under visual distribution shifts with onboard inference. These results show that structured re-noising enables cross-horizon revision and improves performance across robotic sequential prediction tasks.