DOJO
a differentiable simulator for robotics
Taylor Howell & Simon Le Cleac'h
Zico Kolter, Mac Schwager, Zachary Manchester
Abstract
We present a differentiable rigid-body-dynamics simulator for robotics that prioritizes physical accuracy and differentiability: Dojo. The simulator utilizes an expressive maximal-coordinates representation, achieves stable simulation at low sample rates, and conserves energy and momentum by employing a variational integrator. A nonlinear complementarity problem, with nonlinear friction cones, models hard contact and is reliably solved using a custom primal-dual interior-point method. The implicit-function theorem enables efficient differentiation of an intermediate relaxed problem and computes smooth gradients from the contact model. We demonstrate the usefulness of the simulator and it’s gradients through a number of examples including: simulation, trajectory optimization, reinforcement learning, and system identification.
Links
arXiv preprint: https://arxiv.org/abs/2203.00806
Python interface: https://github.com/dojo-sim/dojopy
video presentation: https://youtu.be/TRtOESXJxJQ
BiBteX
@article{howelllecleach2022,
title={Dojo: A Differentiable Simulator for Robotics},
author={Taylor, A. Howell and Le Cleac'h, Simon and Kolter, Zico and Schwager, Mac and Manchester, Zachary},
journal={arXiv preprint arXiv:2203.00806},
year={2022}
}
Features
Differentiable
Non-linear friction cone
Energy & momentum conservation
Application domains
Reinforcement Learning
Trajectory Optimization
System Identification
Environments
Real hardware
Gym
Classic control
pendulum
cartpole
Physics Simulation
Acknowledgments
The authors would like to thank Jan Brüdigam for his contributions to the open-source libraries ConstrainedDynamics.jl and GraphBasedSystems.jl which served as a foundation for Dojo, as well as early technical discussions and support; and Suvansh Sanjeev for assistance with the Python interface. Toyota Research Institute provided funds to support this work.
Authors