动态量化让机器人视觉模型更省资源、跑得更快
DyQ-VLA: Temporal-Dynamic-Aware Quantization for Embodied Vision-Language-Action Models
- 根据动作实时敏感度自动调节精度，避免固定精度浪费
- 仅需原内存30.9%就保持99.5%性能，仿真提速1.49倍
- 适合部署在边缘设备的机器人视觉系统，提升响应速度
视觉-语言-动作（VLA）模型在具身智能中占据主导地位，但推理开销限制了其在边缘设备的应用。尽管模型量化可缓解瓶颈，但静态量化对VLAs效果不佳，主要面临两大挑战：(1) 时间动态敏感性，即固定精度无法适应不同阶段的容错差异；(2) 实时比特分配难题，难以识别实时敏感性以指导位宽分配。为此，我们提出DyQ-VLA，一种面向具身VLA的动态量化框架。该框架采用敏感度感知的切换策略，利用实时运动学代理触发位宽切换，并通过运动学引导模块动态分配最优位宽。实验表明，DyQ-VLA仅需原模型30.9%的内存占用，即可保持99.5%的原始性能，在仿真中实现1.49倍加速，真实场景下最高达1.43倍加速。
原文摘要 · Abstract (English)
Vision-Language-Action (VLA) models are dominant in embodied intelligence but are constrained by inference overheads. While model quantization alleviates these bottlenecks for edge deployment, static quantization approaches remain suboptimal for VLAs due to two critical challenges: (1) Temporal-dynamic sensitivity, where fixed precision wastes resources by ignoring stage-varying error tolerances; and (2) Real-time allocation, where identifying real-time sensitivity to guide bit allocation remains unsolved. To address these challenges, we propose DyQ-VLA, a dynamic quantization framework for VLAs. Specifically, a sensitivity-aware switching strategy leverages real-time kinematic proxies to trigger the bit-width switch, while a kinematic-guided module dynamically allocates the optimal bit-width. Experiments show that DyQ-VLA requires only 30.9% of the original memory footprint while maintaining 99.5% of its original performance, achieving 1.49x simulation and up to 1.43x real-world speedups.
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。