Octopi-X: Robotic Perception with a Large Tactile-Vision-Language Model for Physical Property Inference

Abstract

Inferring physical properties can significantly enhance robotic manipulation by enabling robots to handle objects safely and efficiently through adaptive grasping strategies. Previous approaches have typically relied on either tactile or visual data, limiting their ability to fully capture properties. We introduce a novel cross-modal perception framework that integrates visual observations with tactile representations within a multimodal vision-language model. Our physical reasoning framework, which employs a hierarchical feature alignment mechanism and a refined prompting strategy, enables our model to make property-specific predictions that strongly correlate with ground-truth measurements. Evaluated on 35 diverse objects, our approach outperforms existing baselines and demonstrates strong zero-shot generalization.

Acknowledgment

This research is supported by "Natural Science Foundation of Top Talent of SZTU" (Grant No. GDRC202411)

Citation

@inproceedings{chen2025octopi-x,
title     = {Octopi-X: Robotic Perception with a Large Tactile-Vision-Language Model for Physical Property Inference},
author    = {Zexiang Guo, Hengxiang Chen, Xinheng Mai, Qiusang Qiu, Gan Ma, Zhanat Kappassov, Qiang Li, and Nutan Chen},
booktitle = {IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) Workshop},
year      = {2025},
url       = {https://openreview.net/forum?id=fx00MuY7RW}
}