TaF-Dataset is a large-scale, synchronized tactile-force dataset introduced in
TaF-VLA: Tactile-Force Alignment in Vision-Language-Action Models for
Force-aware Manipulation. It is designed for learning force-aware tactile
representations and studying contact-rich robotic manipulation.
During data collection, an ATI six-axis force/torque sensor records the
interaction wrench (Fx, Fy, Fz, Tx, Ty, and Tz). A piezoelectric
sensing array provides a spatial pressure map… See the full description on the dataset page:
https://huggingface.co/datasets/jiamig/taf-dataset.
