React When You Need To: Event-Triggered Asynchronous Inference for VLA Policies

2026-09-18

AI Summary

本文提出了一种事件触发的动态推理策略，根据场景变化自适应调整推理间隔，以提高VLA模型对环境变化的反应能力。

Abstract

Vision-Language-Action (VLA) models commonly predict action chunks, limiting their ability to react to environmental changes during execution. Existing asynchronous inference methods improve reactivity but typically rely on a fixed inference gap. In this paper, we propose an event-guided dynamic inference strategy that adapts the inference gap according to scene changes observed since the previous inference. Thereby, it simultaneously preserves motion consistency and prompt reactivity. Across static and dynamic real-world settings, our method consistently performs best, averaging 95% success and exceeding the strongest baseline by 55 percentage points. The code will be made publicly available upon acceptance. The project page is available at https://react-when-you-need-to.github.io/.

Problem

• Vision-Language-Action
• asynchronous inference
• environmental changes
• inference gap
• reactivity

Innovation

• event-guided dynamic inference
• adaptive inference gap
• motion consistency
• prompt reactivity