Reducing Temporal Redundancy for Efficient Vision-Language-Action Inference
arXiv:2607.12287
The paper introduces a system-level acceleration for vision-language-action models by incrementally updating visual tokens for dynamic regions and compressing diffusion-based policy sampling into a two-step schedule, achieving over 2× faster inference while keeping high manipulation success rates.
Abstract
Vision-Language-Action (VLA) models exhibit strong generalization for robotic manipulation, yet their high inference latency limits real time deployment. We identify two primary sources of temporal redundancy in existing VLA pipelines: repeated visual encoding of highly similar consecutive frames and multi step iterative sampling in diffusion based policies. To address this, we propose a system level acceleration strategy that reduces computation in both perception and action generation. On the perception side, we incrementally update only tokens corresponding to dynamic scene regions instead of re-encoding entire frames. On the policy side, we compress diffusion sampling into a compact 2-step schedule through efficiency oriented training while preserving action precision. Experiments on Libero, RobotWin, and Real Robot Platforms demonstrate over 2 times speedup while maintaining high performance, achieving up to 98% success rate on general manipulation benchmarks. Our codes will be released on Github.
13pages, 7 figuers