GPT-6-Astra Understands Touch and Outperform Vision--Tactile--Language--Action Models

Ziyong Wang; Yupei Lin; Runxuan Huang; Liang Lin; Dongyu Zhang; Guangrun Wang

doi:10.48550/arXiv.2610.28457

← Back to publications

2026 · Preprint

GPT-6-Astra Understands Touch and Outperform Vision--Tactile--Language--Action Models

Ziyong Wang, Yupei Lin, Runxuan Huang, Liang Lin, Dongyu Zhang, Guangrun Wang

arXiv:2610.28457

Figure 1 — GPT-6-Astra Understands Touch and Outperform Vision--Tactile--Language--Action Models — Figure 1: GPT-as-Tactile-Policy for contact-rich manipulation.

Overview

Investigates whether a general-purpose multimodal model can interpret tactile feedback and use it to guide contact-rich manipulation, combining expert demonstrations, tactile references, multimodal observations, and interaction history.

Paper / Publisher ↗ PDF ↗ Code / Project ↗