Learning to Feel the Future: DreamTacVLA for Contact-Rich Manipulation

Guo Ye, Zexi Zhang, Xu Zhao +4

Vision-Language-Action (VLA) models have shown remarkable generalization by mapping web-scale knowledge to robotic control, yet they remain blind to physical contact. Consequently,…