← Back to Robotics
cs.RO

Why robots need to feel: new dataset teaches touch-aware manipulation

Amirhosein Alian, Yongqiang Zhao, Shiyi Gu, Xuyang Zhang, Zhuo Chen, Christopher E. Mower, Haitham Bou-Ammar, Shan Luo

June 3, 2026

Most robot learning datasets rely only on vision, missing critical tactile signals that prevent breaking eggs and folding cloth. HapTile adds fingerprint-level touch sensors to robots and haptic feedback gloves to human operators, capturing real contact dynamics during pick-and-place, folding, and stacking. Benchmarks show tactile information improves policy learning on contact-rich tasks.
Published as HapTile: A Haptic-Informed Vision-Tactile-Language-Action Dataset for Contact-Rich Imitation Learning arXiv:2606.04825
Read the original paper →