Deep Object-Centric Policies for Autonomous Driving
This addresses robustness and interpretability issues in autonomous driving for robotics applications, though it is incremental as it builds on existing object-centric and end-to-end learning approaches.
The paper tackles the problem of uninterpretable and unreliable end-to-end visuomotor learning in autonomous driving by proposing object-centric models that explicitly represent objects, showing they outperform object-agnostic methods in simulated scenes with vehicles and pedestrians and in real-world low-data regimes.
While learning visuomotor skills in an end-to-end manner is appealing, deep neural networks are often uninterpretable and fail in surprising ways. 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.