通过减少视觉与动作生成中的冗余计算，实现机器人操作的高效推理
Reducing Temporal Redundancy for Efficient Vision-Language-Action Inference
- 仅更新动态场景区域的视觉编码，避免重复处理相似帧
- 将扩散采样压缩为两步，速度提升超2倍且成功率仍达98%
- 适合需要实时响应的机器人控制场景
视觉-语言-动作（VLA）模型在机器人操作中表现出强大泛化能力，但高推理延迟限制了实时部署。我们识别出现有VLA流程中的两大时间冗余来源：连续帧间高度相似的视觉编码重复以及基于扩散模型策略的多步迭代采样。为此，提出系统级加速策略，在感知和动作生成两方面减少计算开销。感知端采用增量更新机制，仅对动态场景区域对应的视觉令牌进行更新，避免完整帧重新编码；策略端通过面向效率的训练，将扩散采样压缩为紧凑的两步调度，同时保持动作精度。在Libero、RobotWin及真实机器人平台上的实验表明，该方法实现超过2倍的速度提升，且在通用操作基准上最高达成98%的成功率。代码将开源于Github。
原文摘要 · Abstract (English)
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.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。