Report: Provable safe reinforcement learning

Summary
Report Provable safe reinforcement learning TUG