This paper proposes a data-driven charging and safety protocol design approach to address the tradeoff between charging speed and aging in lithium-ion batteries. Using a high-fidelity, physics-based battery model, we propose a hybrid control strategy that combines reinforcement learning (RL) and data-driven formal methods via counterexample-guided inductive synthesis. RL synthesizes individual controllers and, through data-driven abstraction, decomposes them into a structure that switches controllers based on initial battery output measurements. We implement a hybrid system by combining discrete selection between RL-based controllers with continuous battery dynamics. Once the design satisfies the requirements, abstraction provides probabilistic guarantees on closed-loop performance.
