Bridging control and reinforcement learning with partial model knowledge