Perfect Demo Makes Poor Teacher: Learning Robust Alignment from Critical Motion Segments
Fluent expert demos rush through alignment and recovery. Slowing down there and resampling those segments helps; STAIR adds a compact short-horizon motion feature between the VLM and the action expert.熟练的示教会把对准与纠错一带而过，导致策略在关键处缺少监督。在这些片段放慢、重采样有帮助；STAIR 进一步在 VLM 与动作专家之间加入紧凑的短时运动特征。
BibTeX
@misc{liu2026perfect,
title = {{Perfect Demo Makes Poor Teacher: Learning Robust Alignment from Critical Motion Segments}},
author = {Mingyu Liu and Zeju Li and Jiuhe Shu and Hanqing Wang and Yuhao Chao and Hao Chen and Chunhua Shen},
year = {2026},
eprint = {2606.15587},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2606.15587},
}