ForceTwin: Physics-informed Digital Twins for Robotic Manipulation from Instrumented Human Interaction

ForceTwin introduces a framework for building physics-informed digital twins of articulated objects through instrumented human interaction. The method learns physical properties such as inertia, friction, and spring mechanisms that determine manipulation difficulty but are not directly observable from visual appearance. These learned properties enable the digital twin to predict object behavior during robotic manipulation, addressing the challenge that visually identical objects may require very different effort to operate.