Dynamic task allocations with Q-learning based particle swarm optimization for human-robot collaboration disassembly of electric vehicle battery recycling

With the wide application of Electric Vehicles (EV) and intelligent technology, the recycling of EV batteries presents significant challenges to cope with the dynamic disassembly tasks and operations. Similarly, human-robot collaborative (HRC) disassembly has emerged as an effective solution to accomplish the disassembly task allocation and improve efficiency. This paper proposed an improved HRC disassembly method integrating Q-learning-based Particle Swarm Optimization (PSO) to optimize the disassembly task sequence for multi-agent disassembly strategies. Furthermore, the Q-learning model is integrated to guide the variable neighborhood search (VNS) algorithm enabling efficient neighborhood structure selection to enhance local search capabilities and optimize multi-agent disassembly sorting tasks. By considering the disassembly experiment of Mercedes-Benz EQS NCM 811 battery as a case study, the proposed method is deeply analyzed and compared with traditional methods under the same objective function. The results demonstrate the proposed algorithm reduces the fitness value and improves the optimization of disassembly tasks. The experimental results show that the proposed algorithm effectively improves the efficiency of HRC disassembly compared to other algorithms, offering a more efficient solution for the recycling of retired EV batteries.