FlashVLA：流式动作解码加速异步 VLA 推理
原标题：FlashVLA: Streaming Action Decoding for Fast and Asynchronous VLA Inference
产业动态AI 82
来源：arXiv cs.RO发布时间待核实
arXiv:2608.27384v1 Announce Type: new Abstract: Vision-Language-Action (VLA) models are increasingly promising for robotic manipulation, yet their real-world deployment remains bottlenecked by high inference latency and unstable asynchronous execution. This challenge is particularly pronounced in flow-matching-based VLA models, where action decoding requires multiple iterative steps conditioned on the VLM context.