Driving SMARTS Competition at NeurIPS 2022: Insights and Outcome

Amir Rasouli, Soheil Alizadeh, Iuliia Kotseruba, Yi Ma, Hebin Liang, Yuan Tian, Zhiyu Huang, Haochen Liu, Jingda Wu, Randy Goebel, Tianpei Yang, Matthew E. Taylor, Liam Paull, Xi Chen
Proceedings of the NeurIPS 2022 Competitions Track, PMLR 220:73-84, 2022.

Abstract

The Driving SMARTS (Scalable Multi-Agent Reinforcement Learning Training School) competition was designed to address one of the major challenges for autonomous driving (AD), namely adaptation to distribution shift between data used for training and inference and the problems caused by this shift in real-world conditions. The two key features of the competition are 1) a two-track structure to encourage and support a variety of approaches to solving the problem, such as reinforcement learning, offline learning, and other machine learning methods; and 2) curated data for driving scenarios of varying difficulty levels, from cruising to unprotected turns at unsignalized intersections. The competition attracted 87 participants in 53 teams. Top-ranking teams contributed a diverse set of solutions highlighting the effectiveness of different methodologies on safe motion planning for AD. This paper provides an overview of the Driving SMARTS competition, discusses its organisational and design aspects, and presents the results, insights, and promising directions for future research.

Cite this Paper

BibTeX
@InProceedings{pmlr-v220-rasouli23a, title = {Driving SMARTS Competition at NeurIPS 2022: Insights and Outcome}, author = {Rasouli, Amir and Alizadeh, Soheil and Kotseruba, Iuliia and Ma, Yi and Liang, Hebin and Tian, Yuan and Huang, Zhiyu and Liu, Haochen and Wu, Jingda and Goebel, Randy and Yang, Tianpei and Taylor, Matthew E. and Paull, Liam and Chen, Xi}, booktitle = {Proceedings of the NeurIPS 2022 Competitions Track}, pages = {73--84}, year = {2022}, editor = {Ciccone, Marco and Stolovitzky, Gustavo and Albrecht, Jacob}, volume = {220}, series = {Proceedings of Machine Learning Research}, month = {28 Nov--09 Dec}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v220/rasouli23a/rasouli23a.pdf}, url = {https://proceedings.mlr.press/v220/rasouli23a.html}, abstract = {The Driving SMARTS (Scalable Multi-Agent Reinforcement Learning Training School) competition was designed to address one of the major challenges for autonomous driving (AD), namely adaptation to distribution shift between data used for training and inference and the problems caused by this shift in real-world conditions. The two key features of the competition are 1) a two-track structure to encourage and support a variety of approaches to solving the problem, such as reinforcement learning, offline learning, and other machine learning methods; and 2) curated data for driving scenarios of varying difficulty levels, from cruising to unprotected turns at unsignalized intersections. The competition attracted 87 participants in 53 teams. Top-ranking teams contributed a diverse set of solutions highlighting the effectiveness of different methodologies on safe motion planning for AD. This paper provides an overview of the Driving SMARTS competition, discusses its organisational and design aspects, and presents the results, insights, and promising directions for future research.} }
Endnote
%0 Conference Paper %T Driving SMARTS Competition at NeurIPS 2022: Insights and Outcome %A Amir Rasouli %A Soheil Alizadeh %A Iuliia Kotseruba %A Yi Ma %A Hebin Liang %A Yuan Tian %A Zhiyu Huang %A Haochen Liu %A Jingda Wu %A Randy Goebel %A Tianpei Yang %A Matthew E. Taylor %A Liam Paull %A Xi Chen %B Proceedings of the NeurIPS 2022 Competitions Track %C Proceedings of Machine Learning Research %D 2022 %E Marco Ciccone %E Gustavo Stolovitzky %E Jacob Albrecht %F pmlr-v220-rasouli23a %I PMLR %P 73--84 %U https://proceedings.mlr.press/v220/rasouli23a.html %V 220 %X The Driving SMARTS (Scalable Multi-Agent Reinforcement Learning Training School) competition was designed to address one of the major challenges for autonomous driving (AD), namely adaptation to distribution shift between data used for training and inference and the problems caused by this shift in real-world conditions. The two key features of the competition are 1) a two-track structure to encourage and support a variety of approaches to solving the problem, such as reinforcement learning, offline learning, and other machine learning methods; and 2) curated data for driving scenarios of varying difficulty levels, from cruising to unprotected turns at unsignalized intersections. The competition attracted 87 participants in 53 teams. Top-ranking teams contributed a diverse set of solutions highlighting the effectiveness of different methodologies on safe motion planning for AD. This paper provides an overview of the Driving SMARTS competition, discusses its organisational and design aspects, and presents the results, insights, and promising directions for future research.
APA
Rasouli, A., Alizadeh, S., Kotseruba, I., Ma, Y., Liang, H., Tian, Y., Huang, Z., Liu, H., Wu, J., Goebel, R., Yang, T., Taylor, M.E., Paull, L. & Chen, X.. (2022). Driving SMARTS Competition at NeurIPS 2022: Insights and Outcome. Proceedings of the NeurIPS 2022 Competitions Track, in Proceedings of Machine Learning Research 220:73-84 Available from https://proceedings.mlr.press/v220/rasouli23a.html.

Related Material