Abstract: Reinforcement learning (RL) holds great potential for robotic skill acquisition, but its practical deployment in industrial disassembly tasks is challenged by low sample efficiency and ...