CAT: Closed-loop Adversarial Training for Safe End-to-End Driving

TL;DR

This paper introduces a Closed-loop Adversarial Training framework for autonomous driving that dynamically generates safety-critical scenarios to improve driving safety and robustness efficiently.

Contribution

The paper proposes a novel environment augmentation method using probabilistic resampling for adversarial scenario generation in end-to-end driving.

Findings

CAT generates effective adversarial scenarios that challenge the driving agent.

Agents trained with CAT show improved safety performance in diverse scenarios.

The approach reduces computational costs compared to existing methods.

Abstract

Driving safety is a top priority for autonomous vehicles. Orthogonal to prior work handling accident-prone traffic events by algorithm designs at the policy level, we investigate a Closed-loop Adversarial Training (CAT) framework for safe end-to-end driving in this paper through the lens of environment augmentation. CAT aims to continuously improve the safety of driving agents by training the agent on safety-critical scenarios that are dynamically generated over time. A novel resampling technique is developed to turn log-replay real-world driving scenarios into safety-critical ones via probabilistic factorization, where the adversarial traffic generation is modeled as the multiplication of standard motion prediction sub-problems. Consequently, CAT can launch more efficient physical attacks compared to existing safety-critical scenario generation methods and yields a significantly less computational cost in the iterative learning pipeline. We incorporate CAT into the MetaDrive simulator and validate our approach on hundreds of driving scenarios imported from real-world driving datasets. Experimental results demonstrate that CAT can effectively generate adversarial scenarios countering the agent being trained. After training, the agent can achieve superior driving safety in both log-replay and safety-critical traffic scenarios on the held-out test set. Code and data are available at https://metadriverse.github.io/cat.