DREAM-Chunk boosts robot action robustness with latent world model

Action chunking has become a common interface for vision-language-action (VLA) models, enabling low-frequency policy inference to drive high-frequency robot execution. However, once an action chunk is committed, its open-loop execution can be brittle under stochastic dynamics, hardware errors, and partial observability. DREAM-Chunk addresses this by adding a lightweight latent world model at test time, without requiring additional policy fine-tuning. It samples multiple candidate action chunks, rolls out their predicted latent futures, and selects actions from the chunk whose predicted state best matches the observed rollout. This uses extra test-time computation to cover multiple plausible stochastic futures and improve reactivity during long-horizon chunk execution.

On the Kinetix benchmark, DREAM-Chunk improves robustness under increasing action noise and benefits from larger candidate sample sizes, especially when demonstrations contain corrective behaviors. The method was further validated on four manipulation tasks across two robot platforms and two VLA policies under various sources of stochasticity. Both simulation and hardware experiments show that DREAM-Chunk consistently improves the robustness of action-chunking policies in stochastic dynamics, making it a practical plug-in enhancement for existing robotic systems.