Comprehensive Training and Evaluation on Deep Reinforcement Learning for Automated Driving in Various Simulated Driving Maneuvers

Comprehensive Training and Evaluation on Deep Reinforcement Learning for Automated Driving in Various Simulated Driving Maneuvers

Dong, Y. (TU Delft Transport and Planning)
Datema, T. (TU Delft Electrical Engineering, Mathematics and Computer Science)
Wassenaar, V. (TU Delft Electrical Engineering, Mathematics and Computer Science)
van de Weg, J.J. (TU Delft Electrical Engineering, Mathematics and Computer Science)
Kopar, C.T. (TU Delft Mechanical Engineering)
Suleman, H.I. (TU Delft Electrical Engineering, Mathematics and Computer Science)

Electrical Engineering, Mathematics and Computer Science

2023

Developing and testing automated driving models in the real world might be challenging and even dangerous, while simulation can help with this, especially for challenging maneuvers. Deep reinforcement learning (DRL) has the potential to tackle complex decision-making and controlling tasks through learning and interacting with the environment, thus it is suitable for developing automated driving while not being explored in detail yet. This study carried out a comprehensive study by implementing, evaluating, and comparing the two DRL algorithms, Deep Q-networks (DQN) and Trust Region Policy Optimization (TRPO), for training automated driving on the highway-env simulation platform. Effective and customized reward functions were developed and the implemented algorithms were evaluated in terms of onlane accuracy (how well the car drives on the road within the lane), efficiency (how fast the car drives), safety (how likely the car is to crash into obstacles), and comfort (how much the car makes jerks, e.g., suddenly accelerates or brakes). Results show that the TRPO-based models with modified reward functions delivered the best performance in most cases. Furthermore, to train a uniform driving model that can tackle various driving maneuvers besides the specific ones, this study expanded the highway-env and developed an extra customized training environment, namely, ComplexRoads, integrating various driving maneuvers and multiple road scenarios together. Models trained on the designed ComplexRoads environment can adapt well to other driving maneuvers with promising overall performance. Lastly, several functionalities were added to the highway-env to implement this work. The codes are open on GitHub at https://github.com/alaineman/drlcarsim-paper.

http://resolver.tudelft.nl/uuid:7262e4d4-2415-4c39-a326-e0488b2b01d5

https://doi.org/10.1109/ITSC57777.2023.10422159

IEEE

2024-03-28

9798350399462

2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC)

26th IEEE International Conference on Intelligent Transportation Systems, ITSC 2023, 2023-09-24 → 2023-09-28, Euskalduna Conference Centre, Bilbao, Spain

IEEE Conference on Intelligent Transportation Systems, Proceedings, ITSC, 2153-0009

Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

Institutional Repository

conference paper

© 2023 Y. Dong, T. Datema, V. Wassenaar, J.J. van de Weg, C.T. Kopar, H.I. Suleman