ViPRA: Video Prediction for Robot Actions

TL;DR

ViPRA introduces a novel framework that leverages actionless videos to learn continuous robot control by predicting scene dynamics and latent actions, enabling effective transfer to real-world manipulation tasks.

Contribution

It presents a pretraining-finetuning approach that learns physically grounded latent actions from unlabeled videos, supporting high-frequency control and cross-embodiment generalization.

Findings

Achieves 16% improvement on the SIMPLER benchmark.

Enables smooth control at up to 22 Hz.

Outperforms prior latent action methods in real-world tasks.

Abstract

Can we turn a video prediction model into a robot policy? Videos, including those of humans or teleoperated robots, capture rich physical interactions. However, most of them lack labeled actions, which limits their use in robot learning. We present Video Prediction for Robot Actions (ViPRA), a simple pretraining-finetuning framework that learns continuous robot control from these actionless videos. Instead of directly predicting actions, we train a video-language model to predict both future visual observations and motion-centric latent actions, which serve as intermediate representations of scene dynamics. We train these latent actions using perceptual losses and optical flow consistency to ensure they reflect physically grounded behavior. For downstream control, we introduce a chunked flow matching decoder that maps latent actions to robot-specific continuous action sequences, using only 100 to 200 teleoperated demonstrations. This approach avoids expensive action annotation, supports generalization across embodiments, and enables smooth, high-frequency continuous control upto 22 Hz via chunked action decoding. Unlike prior latent action works that treat pretraining as autoregressive policy learning, ViPRA explicitly models both what changes and how. Our method outperforms strong baselines, with a 16% gain on the SIMPLER benchmark and a 13% improvement across real world manipulation tasks. We have released models and code at https://vipra-project.github.io