AI RESEARCH
HeiSD: Hybrid Speculative Decoding for Embodied Vision-Language-Action Models with Kinematic Awareness
arXiv CS.LG
•
ArXi:2603.17573v1 Announce Type: cross Vision-Language-Action (VLA) Models have become the mainstream solution for robot control, but suffer from slow inference speeds. Speculative Decoding (SD) is a promising acceleration method which can be divided into two categories: drafter-based SD and retrieval-based SD. Existing methods fail to analyze the advantages and disadvantages of these two types of SD in VLA models, leading to their sole application or optimization.