Machine learning approaches often lack safety guarantees, which are often a key requirement in real-world tasks. This paper addresses the lack of safety guarantees by extending reinforcement learning with a safety layer that restricts the action space to the subspace of safe actions. We demonstrate the proposed approach using lane changing in autonomous driving. To distinguish safe actions from unsafe ones, we compare planned motions with the set of possible occupancies of traffic participants generated by set-based predictions. In situations where no safe action exists, a verified failsafe controller is executed. We used real-world highway traffic data to train and test the proposed approach. The evaluation result shows that the proposed approach trains agents that do not cause collisions during training and deployment.

No abstract

Ensuring safety of reinforcement learning (RL) algorithms is crucial for many real-world tasks. However, vanilla RL does not guarantee safety for an agent. In recent years, several methods have been proposed to provide safety guarantees for RL. To the best of our knowledge, there is no comprehensive comparison of these provably safe RL methods. We therefore introduce a categorization for existing provably safe RL methods, and present the theoretical foundations for both continuous and discrete action spaces. Additionally, we evaluate provably safe RL on an inverted pendulum. In the experiments, it is shown that indeed only provably safe RL methods guarantee safety.

No abstract

No abstract

While reinforcement learning produces very promising results for many applications, its main disadvantage is the lack of safety guarantees, which prevents its use in safety-critical systems. In this work, we address this issue by a safety shield for nonlinear continuous systems that solve reach-avoid tasks. Our safety shield prevents applying potentially unsafe actions from a reinforcement learning agent by projecting the proposed action to the closest safe action. This approach is called action projection and is implemented via mixed-integer optimization. The safety constraints for action projection are obtained by applying parameterized reachability analysis using polynomial zonotopes, which enables to accurately capture the nonlinear effects of the actions on the system. In contrast to other state-of-the-art approaches for action projection, our safety shield can efficiently handle input constraints and dynamic obstacles, eases incorporation of the spatial robot dimensions into the safety constraints, guarantees robust safety despite process noise and measurement errors, and is well suited for high-dimensional systems, as we demonstrate on several challenging benchmark systems.

No abstract

No abstract