Driving Policy Transfer via Modularity and Abstraction

Matthias Mueller, Alexey Dosovitskiy, Bernard Ghanem, Vladlen Koltun
Proceedings of The 2nd Conference on Robot Learning, PMLR 87:1-15, 2018.

Abstract

End-to-end approaches to autonomous driving have high sample complexity and are difficult to scale to realistic urban driving. Simulation can help end-to-end driving systems by providing a cheap, safe, and diverse training environment. Yet training driving policies in simulation brings up the problem of transferring such policies to the real world. We present an approach to transferring driving policies from simulation to reality via modularity and abstraction. Our approach is inspired by classic driving systems and aims to combine the benefits of modular architectures and end-to-end deep learning approaches. The key idea is to encapsulate the driving policy such that it is not directly exposed to raw perceptual input or low-level vehicle dynamics. We evaluate the presented approach in simulated urban environments and in the real world. In particular, we transfer a driving policy trained in simulation to a 1/5-scale robotic truck that is deployed in a variety of conditions, with no finetuning, on two continents.

Cite this Paper


BibTeX
@InProceedings{pmlr-v87-mueller18a, title = {Driving Policy Transfer via Modularity and Abstraction}, author = {Mueller, Matthias and Dosovitskiy, Alexey and Ghanem, Bernard and Koltun, Vladlen}, booktitle = {Proceedings of The 2nd Conference on Robot Learning}, pages = {1--15}, year = {2018}, editor = {Billard, Aude and Dragan, Anca and Peters, Jan and Morimoto, Jun}, volume = {87}, series = {Proceedings of Machine Learning Research}, month = {29--31 Oct}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v87/mueller18a/mueller18a.pdf}, url = {https://proceedings.mlr.press/v87/mueller18a.html}, abstract = {End-to-end approaches to autonomous driving have high sample complexity and are difficult to scale to realistic urban driving. Simulation can help end-to-end driving systems by providing a cheap, safe, and diverse training environment. Yet training driving policies in simulation brings up the problem of transferring such policies to the real world. We present an approach to transferring driving policies from simulation to reality via modularity and abstraction. Our approach is inspired by classic driving systems and aims to combine the benefits of modular architectures and end-to-end deep learning approaches. The key idea is to encapsulate the driving policy such that it is not directly exposed to raw perceptual input or low-level vehicle dynamics. We evaluate the presented approach in simulated urban environments and in the real world. In particular, we transfer a driving policy trained in simulation to a 1/5-scale robotic truck that is deployed in a variety of conditions, with no finetuning, on two continents.} }
Endnote
%0 Conference Paper %T Driving Policy Transfer via Modularity and Abstraction %A Matthias Mueller %A Alexey Dosovitskiy %A Bernard Ghanem %A Vladlen Koltun %B Proceedings of The 2nd Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2018 %E Aude Billard %E Anca Dragan %E Jan Peters %E Jun Morimoto %F pmlr-v87-mueller18a %I PMLR %P 1--15 %U https://proceedings.mlr.press/v87/mueller18a.html %V 87 %X End-to-end approaches to autonomous driving have high sample complexity and are difficult to scale to realistic urban driving. Simulation can help end-to-end driving systems by providing a cheap, safe, and diverse training environment. Yet training driving policies in simulation brings up the problem of transferring such policies to the real world. We present an approach to transferring driving policies from simulation to reality via modularity and abstraction. Our approach is inspired by classic driving systems and aims to combine the benefits of modular architectures and end-to-end deep learning approaches. The key idea is to encapsulate the driving policy such that it is not directly exposed to raw perceptual input or low-level vehicle dynamics. We evaluate the presented approach in simulated urban environments and in the real world. In particular, we transfer a driving policy trained in simulation to a 1/5-scale robotic truck that is deployed in a variety of conditions, with no finetuning, on two continents.
APA
Mueller, M., Dosovitskiy, A., Ghanem, B. & Koltun, V.. (2018). Driving Policy Transfer via Modularity and Abstraction. Proceedings of The 2nd Conference on Robot Learning, in Proceedings of Machine Learning Research 87:1-15 Available from https://proceedings.mlr.press/v87/mueller18a.html.

Related Material