PlanNetX: Learning an efficient neural network planner from MPC for longitudinal control

Jasper Hoffmann, Diego Fernandez Clausen, Julien Brosseit, Julian Bernhard, Klemens Esterle, Moritz Werling, Michael Karg, Joschka Joschka Bödecker
Proceedings of the 6th Annual Learning for Dynamics & Control Conference, PMLR 242:1214-1227, 2024.

Abstract

Model predictive control (MPC) is a powerful, optimization-based approach for controlling dynamical systems. However, the computational complexity of online optimization can be problematic on embedded devices. Especially, when we need to guarantee fixed control frequencies. Thus, previous work proposed to reduce the computational burden using imitation learning (IL) approximating the MPC policy by a neural network. In this work, we instead learn the whole planned trajectory of the MPC. We introduce a combination of a novel neural network architecture PlanNetX and a simple loss function based on the state trajectory that leverages the parameterized optimal control structure of the MPC. We validate our approach in the context of autonomous driving by learning a longitudinal planner and benchmarking it extensively in the CommonRoad simulator using synthetic scenarios and scenarios derived from real data. Our experimental results show that we can learn the open-loop MPC trajectory with high accuracy while improving the closed-loop performance of the learned control policy over other baselines like behavior cloning.

Cite this Paper

BibTeX
@InProceedings{pmlr-v242-hoffmann24a, title = {{PlanNetX}: {L}earning an efficient neural network planner from {MPC} for longitudinal control}, author = {Hoffmann, Jasper and Clausen, Diego Fernandez and Brosseit, Julien and Bernhard, Julian and Esterle, Klemens and Werling, Moritz and Karg, Michael and B\"{o}decker, Joschka Joschka}, booktitle = {Proceedings of the 6th Annual Learning for Dynamics & Control Conference}, pages = {1214--1227}, year = {2024}, editor = {Abate, Alessandro and Cannon, Mark and Margellos, Kostas and Papachristodoulou, Antonis}, volume = {242}, series = {Proceedings of Machine Learning Research}, month = {15--17 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v242/hoffmann24a/hoffmann24a.pdf}, url = {https://proceedings.mlr.press/v242/hoffmann24a.html}, abstract = {Model predictive control (MPC) is a powerful, optimization-based approach for controlling dynamical systems. However, the computational complexity of online optimization can be problematic on embedded devices. Especially, when we need to guarantee fixed control frequencies. Thus, previous work proposed to reduce the computational burden using imitation learning (IL) approximating the MPC policy by a neural network. In this work, we instead learn the whole planned trajectory of the MPC. We introduce a combination of a novel neural network architecture PlanNetX and a simple loss function based on the state trajectory that leverages the parameterized optimal control structure of the MPC. We validate our approach in the context of autonomous driving by learning a longitudinal planner and benchmarking it extensively in the CommonRoad simulator using synthetic scenarios and scenarios derived from real data. Our experimental results show that we can learn the open-loop MPC trajectory with high accuracy while improving the closed-loop performance of the learned control policy over other baselines like behavior cloning.} }
Endnote
%0 Conference Paper %T PlanNetX: Learning an efficient neural network planner from MPC for longitudinal control %A Jasper Hoffmann %A Diego Fernandez Clausen %A Julien Brosseit %A Julian Bernhard %A Klemens Esterle %A Moritz Werling %A Michael Karg %A Joschka Joschka Bödecker %B Proceedings of the 6th Annual Learning for Dynamics & Control Conference %C Proceedings of Machine Learning Research %D 2024 %E Alessandro Abate %E Mark Cannon %E Kostas Margellos %E Antonis Papachristodoulou %F pmlr-v242-hoffmann24a %I PMLR %P 1214--1227 %U https://proceedings.mlr.press/v242/hoffmann24a.html %V 242 %X Model predictive control (MPC) is a powerful, optimization-based approach for controlling dynamical systems. However, the computational complexity of online optimization can be problematic on embedded devices. Especially, when we need to guarantee fixed control frequencies. Thus, previous work proposed to reduce the computational burden using imitation learning (IL) approximating the MPC policy by a neural network. In this work, we instead learn the whole planned trajectory of the MPC. We introduce a combination of a novel neural network architecture PlanNetX and a simple loss function based on the state trajectory that leverages the parameterized optimal control structure of the MPC. We validate our approach in the context of autonomous driving by learning a longitudinal planner and benchmarking it extensively in the CommonRoad simulator using synthetic scenarios and scenarios derived from real data. Our experimental results show that we can learn the open-loop MPC trajectory with high accuracy while improving the closed-loop performance of the learned control policy over other baselines like behavior cloning.
APA
Hoffmann, J., Clausen, D.F., Brosseit, J., Bernhard, J., Esterle, K., Werling, M., Karg, M. & Bödecker, J.J.. (2024). PlanNetX: Learning an efficient neural network planner from MPC for longitudinal control. Proceedings of the 6th Annual Learning for Dynamics & Control Conference, in Proceedings of Machine Learning Research 242:1214-1227 Available from https://proceedings.mlr.press/v242/hoffmann24a.html.

Related Material