Deep Object-Centric Policies for Autonomous Driving

TL;DR

This paper introduces object-centric models for autonomous driving that improve robustness and interpretability by explicitly representing objects, outperforming traditional methods in both simulated and real-world environments.

Contribution

The paper presents a taxonomy of object-centric models and demonstrates their superior performance over object-agnostic methods in simulation and real-world datasets.

Findings

Object-centric models outperform object-agnostic methods in GTA V simulations.

Object-centric models perform better in low-data regimes on real-world datasets.

Object-centric models remain effective despite imperfect object detectors.

Abstract

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.