Improving Contact-Rich Robotic Simulation with Generalized Rigid-Body Dynamics Algorithms

Matthew Chignoli, Nicholas Adrian, Sang bae Kim, Patrick Wensing

Keywords: Rigid Body Dynamics, Dynamic Simulation

TL;DR: By generalizing classic dynamics algorithms for rigid-body systems, we lay the foundation for a novel dynamic simulator compatible with contact-rich robots containing complex sub-mechanisms.

Abstract:

We propose a novel approach for generalizing rigid-body dynamics algorithms to handle complex sub-mechanisms that are popular in robotic systems.
The approach is based on the unification of local constraint embedding and spatial vector algebra, which allows for the development of compact and computationally efficient algorithms for simulating contact-rich robots.
We derive the approach intuitively using the Gauss Principle of Least Constraint.
Overall, our approach provides a valuable foundation for simulating and controlling complex robotic systems with traditionally difficult-to-simulate designs such as geared motors, differential drives, and four-bar mechanisms.

Confirmation: I am a student and I would like to be considered for the funding award.