🤖 AI Summary
Spherical representations offer computational efficiency in robotic contact planning but struggle to accurately capture physical characteristics such as contact area, pressure distribution, and frictional moments. To address this limitation, this work proposes the Compliant Spherical Lattice Contact (CSLC) model, which enables distributed contact modeling directly on native spherical representations for the first time. The robot surface is modeled as a lattice of compliant spheres interconnected by anchoring and lateral springs, automatically generating spatially distributed contact patches under load. This approach transcends the conventional point-contact assumption, enabling more physically realistic mechanical responses. Through differentiable geometric modeling and a multi-solver validation framework, experiments demonstrate CSLC’s efficacy across two independent solvers, significantly enhancing contact patch formation and grasp stability.
📝 Abstract
Contact planning in robotics requires models that are both computationally efficient and physically accurate. Sphere-based robot representations satisfy the first requirement by enabling fast collision checking and differentiable geometry, but sacrifice physical accuracy by relying on point contact which cannot capture contact patch area, pressure distributions, rotational stiffness, or frictional moments. We introduce Compliant Sphere Lattice Contact (CSLC), a distributed contact model that operates natively on sphere representations by modeling the robot interface as a compliant lattice of surface spheres connected through anchor and lateral springs. When pressed against an object, the lattice deforms to produce a spatially distributed contact patch that improves the physical accuracy of sphere-based contact. We validate CSLC across two independent solvers and show preliminary results demonstrating contact patch formation and improved grasp stability.