Training-Free Acceleration for Vision-Language-Action Models with Action Caching and Refinement

2026-07-07

AI Summary

This work addresses the high inference latency of vision-language-action (VLA) models caused by iterative denoising and proposes ActionCache—a plug-and-play external caching mechanism. Without requiring any retraining, ActionCache compactly indexes historical intermediate actions using multimodal keys and enables cross-task and cross-episode action reuse through context-similarity-based retrieval, thereby providing warm starts for faster inference. Evaluated on flow-matching-based VLA architectures, ActionCache achieves speedups of up to 11.75× (π₀.₅) in simulation and 34.43× (GR00T-N1.6) in real-world environments while maintaining high task success rates.

Abstract

Vision-Language-Action (VLA) models have emerged as a promising approach for generalizable robotic manipulations. In particular, flow matching-based VLA models have shown remarkable success due to their capability to generate precise and smooth action sequences and capture multimodal distributions. However, the iterative denoising process in the action head acts as a major computational bottleneck, posing a critical challenge for real-time deployment. To address this challenge, we propose ActionCache, a plug-and-play external cache that opportunistically reuses past intermediate actions to warm-start generations from the vicinity of target actions, thereby drastically reducing the inference latency. Specifically, ActionCache stores the intermediate actions with compact multimodal keys, which enables retrieval from similar past contexts across different episodes or even different tasks. Experimental results in simulation and real-world environments demonstrate that ActionCache maintains high task success rates in a low-latency regime, achieving inference acceleration of up to $11.75\times$ and $34.43\times$ for representative flow-based VLA models, $π_{0.5}$ and GR00T-N1.6, respectively.

Problem

• Vision-Language-Action models
• iterative denoising
• computational bottleneck
• real-time deployment
• inference latency

Innovation

• Action Caching
• Training-Free Acceleration
• Vision-Language-Action Models
• Flow Matching
• Inference Latency Reduction