Self-Improving Safety Performance of Reinforcement Learning Based Driving with Black-Box Verification Algorithms

In this work, we propose a self-improving artificial intelligence system for enhancing the safety performance of reinforcement learning (RL) based autonomous driving (AD) agents based on black-box verification methods. RL methods have enjoyed popularity among AD applications in recent years. That being said, existing RL algorithms' performance strongly depends on the diversity of training scenarios. Lack of safety-critical scenarios in the training phase might lead to poor generalization performance in real-world driving applications. We propose a novel framework, where the weaknesses of the training set are explored via black-box verification methods. After the discovery of AD failure scenarios, the training of the RL agent is re-initiated to improve the performance of the previously unsafe scenarios. Simulation results show that the proposed approach efficiently discovers such safety failures in RL-based adaptive cruise control (ACC) applications and significantly reduces the number of vehicle collisions through iterative applications of our method.

DeFIX: Detecting and Fixing Failure Scenarios with Reinforcement Learning in Imitation Learning Based Autonomous Driving

Safely navigating through an urban environment without violating any traffic rules is a crucial performance target for reliable autonomous driving. In this paper, we present a Reinforcement Learning (RL) based methodology to DEtect and FIX (DeFIX) failures of an Imitation Learning (IL) agent by extracting infraction spots and re-constructing mini-scenarios on these infraction areas to train an RL agent for fixing the shortcomings of the IL approach. DeFIX is a continuous learning framework, where extraction of failure scenarios and training of RL agents are executed in an infinite loop. After each new policy is trained and added to the library of policies, a policy classifier method effectively decides on which policy to activate at each step during the evaluation. It is demonstrated that even with only one RL agent trained on failure scenario of an IL agent, DeFIX method is either competitive or does outperform state-of-the-art IL and RL based autonomous urban driving benchmarks. We trained and validated our approach on the most challenging map (Town05) of CARLA simulator which involves complex, realistic, and adversarial driving scenarios. The source code is publicly available at https://github.com/data-and-decision-lab/DeFIX