Abstract: Reinforcement learning (RL) for robotics poses significant consideration of safety during training. However, a major challenge of safe reinforcement learning (SRL) methods is the curse of ...