Fast Contact-Implicit Model-Predictive Control

Fast Contact-Implicit Model-Predictive Control

We present a general approach for controlling robotic systems that make and break contact with their environments. Contact-implicit model-predictive control (CI-MPC) generalizes linear MPC to contact-rich settings by relying on linear complementarity problems (LCP) computed using strategic Taylor approximations about a reference trajectory and retaining non-smooth impact and friction dynamics, allowing the policy to not only reason about contact forces and timing, but also generate entirely new contact mode sequences online. To achieve reliable and fast numerical convergence, we devise a structure-exploiting, path-following solver for the LCP contact dynamics and a custom trajectory optimizer for trajectory-tracking MPC problems. We demonstrate CI-MPC at real-time rates in simulation, and show that it is robust to model mismatch and can respond to disturbances by discovering and exploiting new contact modes across a variety of robotic systems, including a pushbot, hopper, and planar quadruped and biped.

Our implementation and examples can be found here.

Related Papers

2021
July
PDF Fast Contact-Implicit Model-Predictive Control
Simon Le Cleac'h, Taylor Howell, Mac Schwager, and Zac Manchester
arXiv (Submitted)

People

Simon Le Cleac'h
Game-theoretic optimization and optimization through contact
Taylor Howell
Robust feedback motion planning and optimization through contact
Zac Manchester
Assistant Professor
Last updated: 2021-09-27