The job-shop and robot scheduling problem with multi-shelf robots and human intervention

We study the Job-Shop and Robot Scheduling Problem (JSRSP), in which robots are equipped with multiple shelves to carry different jobs, and production operations can only start after the corresponding labour completes parameter-setting tasks. The objective is to minimise the makespan. We formulate a mixed-integer programming (MIP) model, analyze key optimality properties, and develop a tailored Adaptive Large Neighbourhood Search (ALNS) heuristic. Computational experiments on benchmark instances show that the proposed ALNS solves all small instances to optimality and significantly outperforms the best existing exact models on medium and large instances. The algorithm is also effective for the special case with unit robot capacity and no human intervention. To further examine the characteristics of the problem, we conduct extensive sensitivity analyses with respect to robot capacity, travel-to-production time ratios, and labour availability. The results indicate that these factors have a significant impact on the makespan. Finally, a real-world case study from Xuzhou Construction Machinery Group (XCMG) is presented, providing managerial insights.