arXiv 2506.101002025-06-11cs.CV

EfficientVLA：面向视觉-语言-动作模型的免训练加速与压缩

EfficientVLA: Training-Free Acceleration and Compression for Vision-Language-Action Models

School of Artificial Intelligence, Shanghai Jiao Tong University（上海交通大学人工智能学院）；Harbin Institute of Technology（哈尔滨工业大学）；Xi’an Jiaotong University（西安交通大学）；University of Electronic Science and Technology of China（电子科学与技术大学）

Yantai Yang, Yuhao Wang, Zichen Wen, Luo Zhongwei, Chang Zou, Zhipeng Zhang, Chuan Wen, Linfeng Zhang

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

针对视觉-语言-动作模型高计算内存需求的问题，提出免训练加速框架EfficientVLA，通过三种协同策略在CogACT模型上实现1.93倍推理加速且性能损失极小。

摘要：

视觉-语言-动作（Vision-Language-Action, VLA）模型，尤其是基于扩散的架构，在具身智能领域展现出变革性潜力，但因其固有的大量冗余以及推理阶段的冗余带来的高计算与内存需求，发展受到严重阻碍。现有加速工作往往针对孤立的低效问题，这类零散的解决方案通常无法全面解决整个VLA流水线中各异的计算与内存瓶颈，从而限制了实际部署能力。我们提出EfficientVLA，这是一个结构化的免训练推理加速框架，通过协同利用多方面的冗余，系统性地消除这些障碍。EfficientVLA协同整合了三种针对性策略：（1）基于层间冗余分析，从语言模块中剪枝功能不重要的层；（2）通过任务感知策略优化视觉处理通路，选择紧凑且多样的视觉token集合，平衡任务关键性与信息覆盖度；（3）通过策略性地缓存和复用关键中间特征，缓解基于扩散的迭代动作头中的时间计算冗余。我们将该方法应用于标准VLA模型CogACT，在SIMPLER基准上实现了1.93倍的推理加速，浮点运算量（FLOPs）降至原有的28.9%，而成功率仅下降0.6%。

英文摘要：

Vision-Language-Action (VLA) models, particularly diffusion-based architectures, demonstrate transformative potential for embodied intelligence but are severely hampered by high computational and memory demands stemming from extensive inherent and inference-time redundancies. While existing acceleration efforts often target isolated inefficiencies, such piecemeal solutions typically fail to holistically address the varied computational and memory bottlenecks across the entire VLA pipeline, thereby limiting practical deployability. We introduce EfficientVLA, a structured and training-free inference acceleration framework that systematically eliminates these barriers by cohesively exploiting multifaceted redundancies. EfficientVLA synergistically integrates three targeted strategies: (1) pruning of functionally inconsequential layers from the language module, guided by an analysis of inter-layer redundancies; (2) optimizing the visual processing pathway through a task-aware strategy that selects a compact, diverse set of visual tokens, balancing task-criticality with informational coverage; and (3) alleviating temporal computational redundancy within the iterative diffusion-based action head by strategically caching and reusing key intermediate features. We apply our method to a standard VLA model CogACT, yielding a 1.93X inference speedup and reduces FLOPs to 28.9%, with only a 0.6% success rate drop in the SIMPLER benchmark.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。