Physics-Informed Force Prediction with Reactive Surface Following for Granular Material Manipulation on Complex Hidden Surfaces

Zijian Cai, Henrik Andreasson, Yasemin Bekiroglu, Johannes A. Stork

Keywords: Robotics, Reactive Control, Active Perception

Abstract: Granular material manipulation in containers with complex geometries is challenging due to hidden rigid surfaces and uncertain interaction forces. This paper proposes a framework that combines physics-informed force prediction with reactive control. A Gaussian Process model predicts the expected interaction force between the tool and the granular material, which is used as a baseline for an adaptive controller. Experiments show that the proposed method captures the forces trend and enables stable surface following, providing a basis for contact identification and probabilistic surface mapping.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 9