Back to Search
Start Over
ContactNets: Learning Discontinuous Contact Dynamics with Smooth, Implicit Representations
- Publication Year :
- 2020
-
Abstract
- Common methods for learning robot dynamics assume motion is continuous, causing unrealistic model predictions for systems undergoing discontinuous impact and stiction behavior. In this work, we resolve this conflict with a smooth, implicit encoding of the structure inherent to contact-induced discontinuities. Our method, ContactNets, learns parameterizations of inter-body signed distance and contact-frame Jacobians, a representation that is compatible with many simulation, control, and planning environments for robotics. We furthermore circumvent the need to differentiate through stiff or non-smooth dynamics with a novel loss function inspired by the principles of complementarity and maximum dissipation. Our method can predict realistic impact, non-penetration, and stiction when trained on 60 seconds of real-world data.<br />Comment: S.P. and M.H. contributed equally to this work; Accepted to CoRL 2020
- Subjects :
- Computer Science - Robotics
Computer Science - Machine Learning
Subjects
Details
- Database :
- arXiv
- Publication Type :
- Report
- Accession number :
- edsarx.2009.11193
- Document Type :
- Working Paper