This repository is by Brandon Amos, Ivan Dario Jimenez Rodriguez, Jacob Sacks, Byron Boots, and J. Zico Kolter and contains the PyTorch source code to reproduce the experiments in our NIPS 2018 paper Differentiable MPC for End-to-end Planning and Control.
The PyTorch implementation of the fast and differentiable MPC solver we developed for this work is available as a standalone library at locuslab/mpc.pytorch.
If you find this repository helpful in your publications, please consider citing our paper.
@article{amos2018differentiable,
title={{Differentiable MPC for End-to-end Planning and Control}},
author={Brandon Amos and Ivan Jimenez and Jacob Sacks and Byron Boots and J. Zico Kolter},
booktitle={{Advances in Neural Information Processing Systems}},
year={2018}
}
From within the imitation_lqr
directory:
train.py
is the main training script for the experiment
in Section 5.3.From within the imitation_nonconvex
directory:
make_dataset.py
should be run to create a dataset of trajectories
for each environment.il_exp.py
is the main training script for each experiment.run-pendulum-cartpole.sh
runs all of the experiments for the
pendulum and cartpole environments in Section 5.3.run-complex-pendulum.sh
runs all of the experiments for the
non-realizable pendulum environment in Section 5.4.