Randomized Adversarial Imitation Learning for Autonomous Driving

Abstract

With the evolution of various advanced driver assistance system (ADAS) platforms, the design of autonomous driving system is becoming more complex and safety-critical. The autonomous driving system simultaneously activates multiple ADAS functions; and thus it is essential to coordinate various ADAS functions. This paper proposes a randomized adversarial imitation learning (RAIL) method that imitates the coordination of autonomous vehicle equipped with advanced sensors. The RAIL policies are trained through derivative-free optimization for the decision maker that coordinates the proper ADAS functions, e.g., smart cruise control and lane keeping system. Especially, the proposed method is also able to deal with the LIDAR data and makes decisions in complex multi-lane highways and multi-agent environments.

Cite

Text

Shin and Kim. "Randomized Adversarial Imitation Learning for Autonomous Driving." International Joint Conference on Artificial Intelligence, 2019. doi:10.24963/IJCAI.2019/638

Markdown

[Shin and Kim. "Randomized Adversarial Imitation Learning for Autonomous Driving." International Joint Conference on Artificial Intelligence, 2019.](https://mlanthology.org/ijcai/2019/shin2019ijcai-randomized/) doi:10.24963/IJCAI.2019/638

BibTeX

@inproceedings{shin2019ijcai-randomized,
  title     = {{Randomized Adversarial Imitation Learning for Autonomous Driving}},
  author    = {Shin, Myungjae and Kim, Joongheon},
  booktitle = {International Joint Conference on Artificial Intelligence},
  year      = {2019},
  pages     = {4590-4596},
  doi       = {10.24963/IJCAI.2019/638},
  url       = {https://mlanthology.org/ijcai/2019/shin2019ijcai-randomized/}
}