Abstract: With the gradual application of reinforcement learning (RL), safety has emerged as a paramount concern. This article presents a novel data-model hybrid-driven safe RL (SRL) scheme to address ...