Semantic Anchoring for Robotic Action Representations

TL;DR: VLA fine-tuning on limited robot data erodes its inherited semantic structure and undermines generalization. Inspired by mirror neuron theory, we probe this erosion and reveal its correlation with model performance. We introduce a plug-and-play semantic alignment method that consistently improves performance on simulation benchmarks and real-robot tasks.

pick up the grapes and place it on the plate	stack the cups	place the toy bear into the box and close both side flaps	place the sponge in the cabinet and close the door

Language Variation	Novel Object	Visual Distraction	Position Variation	Compositional Tasks

@misc{xu2026semanticanchoringroboticaction,
      title={Semantic Anchoring for Robotic Action Representations},
      author={Yuan Xu and Youheng Shi and Chengyang Li and Wentao Zhu and Yizhou Wang},
      year={2026},
      eprint={2607.13597},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2607.13597},
}

This project builds on openpi, SpatialVLA, and StarVLA. We sincerely appreciate the work their authors have shared with the community.