Autonomous driving has the potential to revolutionize mobility and is hence an active area of research. In practice, the behavior of autonomous vehicles must be acceptable, i.e., efficient, safe, and interpretable. While vanilla reinforcement learning (RL) finds performant behavioral strategies, they are often unsafe and uninterpretable. Safety is introduced through Safe RL approaches, but they still mostly remain uninterpretable as the learned behaviour is jointly optimized for safety and performance without modeling them separately. Interpretable machine learning is rarely applied to RL. This paper proposes SafeDQN, which allows to make the behavior of autonomous vehicles safe and interpretable while still being efficient. SafeDQN offers an understandable, semantic trade-off between the expected risk and the utility of actions while being algorithmically transparent. We show that SafeDQN finds interpretable and safe driving policies for a variety of scenarios and demonstrate how state-of-the-art saliency techniques can help to assess both risk and utility.

Computer scientists at the University of California San Diego have developed a navigation system that will allow robots to better negotiate busy clinical environments. A navigation system developed by University of California, San Diego (UC San Diego) computer scientists aims to improve the ability of robots to navigate busy clinical environments, especially hospital emergency departments. The Safety Critical Deep Q-Network (SafeDQN) navigation system is built around an algorithm that factors in the number of people clustered in a space and the speed and abruptness with which they are moving and directs robots to move around them. The researchers trained the algorithm using a dataset of more than 700 YouTube videos, mainly from documentaries and reality shows. When tested in a simulated environment and compared to other state-of-the-art robotic navigation systems, the researchers determined that SafeDQN found the most efficient and safest paths in all cases.