Shaft-integrated Force Sensing with Transformer-based Dynamics Compensation for Telesurgery

Researchers built a reproducible shaft-integrated force sensing setup for a standard cable-driven surgical tool.

A transformer model using robot state data improved end-effector force estimation, reaching under 6% normalized error.

The approach generalized better than proximal sensing methods and is promising for telesurgery, haptic feedback, and skill assessment.

Its main limitation was sensor saturation when internal cable forces became too large.