OFlow: Injecting Object-Aware Temporal Flow Matching for Robust Robotic Manipulation

TL;DR

OFlow introduces a unified framework that combines temporal prediction and object-aware reasoning in a shared latent space, improving robustness in robotic manipulation tasks.

Contribution

It unifies future scene prediction and object reasoning in a shared semantic latent space, enhancing control robustness under distribution shifts.

Findings

Object-aware foresight improves robustness across multiple benchmarks.

Integrating OFlow into VLA pipelines enhances success in real-world tasks.

The method effectively filters task-irrelevant variations in scene understanding.

Abstract

Robust robotic manipulation requires not only predicting how the scene evolves over time, but also recognizing task-relevant objects in complex scenes. However, existing VLA models face two limitations. They typically act only on the current frame, while future prediction and object-aware reasoning are often learned in separate latent spaces. We propose OFlow (injecting Object-Aware Temporal Flow Matching into VLAs), a framework that addresses both limitations by unifying temporal foresight and object-aware reasoning in a shared semantic latent space. Our method forecasts future latents with temporal flow matching, factorizes them into object-aware representations that emphasize physically relevant cues while filtering task-irrelevant variation, and conditions continuous action generation on these predictions. By integrating OFlow into VLA pipelines, our method enables more reliable control under distribution shifts. Extensive experiments across LIBERO, LIBERO-Plus, MetaWorld, and SimplerEnv benchmarks and real-world tasks demonstrate that object-aware foresight consistently enhances robustness and success.