Reinforcement Learning (RL) is a widely employed machine learning architecture that has been applied to a variety of decision-making problems, from resource management to robot locomotion, from recommendation systems to systems biology, and from traffic control to superhuman-level gaming. However, RL has experienced limited success beyond rigidly controlled or constrained applications, and successful employment of RL in safety-critical scenarios is yet to be achieved. A principal reason for this limitation is the lack of formal approaches to specify requirements as tasks and learning constraints, and to provide guarantees with respect to these requirements and constraints, during and after learning. This line of work addresses these issues by proposing a general framework that leverages the success of RL in learning high-performance controllers, while guaranteeing the satisfaction of given requirements and guiding the learning process within safe configurations.
