arxivcs.RO2026-06-30
UniTacVLA: Unified Tactile Understanding and Prediction in Vision Language Action Models
Xidong Zhang, Yichi Zhang, Jiaxin Shi, Fucai Zhu, Siyu Zhu, Michael Yu Wang, et al.
Vision-language-action (VLA) models have achieved strong performance in many robotic manipulation tasks, yet remain limited in contact-rich dexterous manipulation. To overcome this limitation, recent vision-tactile-language-action (VTLA) methods incorporate tactile sensing into V…