DiffSDFSim: Differentiable Rigid-Body Dynamics With Implicit Shapes
This work addresses a bottleneck in computer vision and robotics for scene understanding with complex object shapes, representing an incremental advancement over existing methods limited to simple or known shapes.
The paper tackles the problem of differentiable physics for objects with complex shapes by representing shapes implicitly using signed distance fields (SDFs), enabling contact point calculation for nonconvex shapes and shape optimization via gradient-based methods, and demonstrates inference of physical parameters like friction coefficients and mass from trajectory and depth image observations in synthetic and real scenarios.
Differentiable physics is a powerful tool in computer vision and robotics for scene understanding and reasoning about interactions. Existing approaches have frequently been limited to objects with simple shape or shapes that are known in advance. In this paper, we propose a novel approach to differentiable physics with frictional contacts which represents object shapes implicitly using signed distance fields (SDFs). Our simulation supports contact point calculation even when the involved shapes are nonconvex. Moreover, we propose ways for differentiating the dynamics for the object shape to facilitate shape optimization using gradient-based methods. In our experiments, we demonstrate that our approach allows for model-based inference of physical parameters such as friction coefficients, mass, forces or shape parameters from trajectory and depth image observations in several challenging synthetic scenarios and a real image sequence.