FASA: Feedback-Aware Sampling Adaptation for Efficient Diffusion-Based VLA Models

2026-09-16

AI Summary

为解决基于扩散的VLA模型在边缘平台实时部署时的高计算和内存访问成本问题，提出FASA框架，通过动态调整采样步骤以适应实际交互反馈，从而提高推理速度。

Abstract

Diffusion-based Vision-Language-Action (VLA) models achieve strong performance in embodied tasks, but their iterative sampling imposes heavy computational and memory-access cost, blocking real-time deployment on edge platforms. Existing acceleration methods either require expensive training (e.g., distillation, flow matching) or degrade perception via statically scheduled pruning and caching, ignoring the dynamic workload variance of robotic interactions. This paper presents FASA (Feedback-Aware Sampling Adaptation), a training-free runtime framework that treats real-time multimodal feedback as a control signal for the denoising pipeline: an interaction-driven range adaptor modulates the global sampling-step budget based on visual and gripper-force feedback, and a proprioception-aware step adaptor pinpoints the optimized step within the adapted range. This co-designed framework allows the underlying hardware architecture to adaptively match the workload demands of different execution phases. Comparative evaluations across several benchmarks show that the inference speed can be increased by up to 1.45$\times$ while maintaining competitive success rates, providing a novel dynamic runtime architecture paradigm for deploying heavy generative embodied AI workloads onto resource-constrained computing platforms.

Problem

• diffusion-based VLA models
• real-time deployment
• computational cost
• memory-access cost
• dynamic workload variance

Innovation

• Feedback-Aware
• Sampling Adaptation
• Real-time Multimodal Feedback
• Denoising Pipeline