Yilun Chen

Robots can now predict not just immediate actions but also the next stages of complex tasks, leading to more efficient manipulation strategies.

RoboInter1.5 reimagines intermediate representations as a powerful interface that enhances robotic reasoning and execution, bridging the gap between low-level actions and high-level world dynamics.

By decoupling visual and motor information during pretraining, FutureVLA unlocks more effective visuomotor prediction for vision-language-action models, boosting performance without modifying downstream architectures.