Abstract

While learning visuomotor skills in an end-to-end manner is appealing, deep neural networks are often uninterpretable and fail in surprising way s. For robotics tasks, such as autonomous driving, models that explicitly represent objects may be more robust to new scenes and provide intuitive visualizations. We describe a taxonomy of “object-centric” models which leverage both object instances and end-to-end learning. In the Grand Theft Auto V simulator, we show that object-centric models outperform object-agnostic methods in scenes with other vehicles and pedestrians, even with an imperfect detector. We also demonstrate that our architectures perform well on real-world environments by evaluating on the Berkeley DeepDrive Video dataset, where an object-centric model outperforms object-agnostic models in the low-data regimes.

Paper

Dequan Wang, Coline Devin, Qi-Zhi Cai, Fisher Yu, Trevor Darrell Deep Object-Centric Policies for Autonomous Driving ICRA 2019

Citation

@inproceedings{wang2018deep,
  title={Deep Object Centric Policies for Autonomous Driving},
  author={Wang, Dequan and Devin, Coline and Cai, Qi-Zhi and Yu, Fisher and Darrell, Trevor},
  booktitle={ICRA},
  year={2019}
}