Abstract: In the e-commerce “rack-to-picker” picking system, how to schedule robots in the system and distribute tasks reasonably determines the operating efficiency and cost of the whole system. Through the analysis of the operation process in“rack-to-picker” picking system, a two-layer planning model for robot number configuration, robot scheduling and robot task assignment is established. The upper model calculates the integer programming model by taking the minimum total cost of the bulk order as the objective function and the robot scheduling as the decision variable. The lower model uses the minimum average idle rate of the robot to complete the tasks as the objective function, and the task assignment as the decision variable. Considering the uncertainties of the walking distance caused by the robot in the process of completing the task due to scheduling, obstacle avoidance and path planning, the robust optimization model is established. The upper layer scheduling results constrain the lowest average idle rate of the lower layer, and the lower layer task assignment result affects the minimum cost of the upper layer. The upper and lower layer results together determine the robot configuration decision. The genetic algorithm is used to solve the model, and the effectiveness of the model is verified by an example simulation.

Key words: “rack-to-picker” picking system, multi-robot task assignment, robot scheduling, robust bi-level programming, genetic algorithm

摘要： 在电商“货到人”拣选系统中,如何调度系统中的机器人并对任务进行合理地分配决定着整个系统的运行效率与成本。分析“货到人”拣选系统作业流程,建立机器人数量配置、机器人调度与机器人任务分配的双层规划模型。上层模型以批量订单完成总成本最小为目标函数,以机器人调度为决策变量,构建整数规划模型;下层模型以机器人完成所有任务的平均空闲率最小为目标函数,以任务分配为决策变量,考虑机器人在完成任务过程中由于调度、避障、路径规划等导致的行走距离不确定因素,构建鲁棒优化模型。上层的调度结果制约了下层的最小平均空闲率,下层的任务分配结果影响上层的最小成本,上下层结果共同决定机器人配置决策。利用遗传算法求解模型,通过实例仿真验证了模型的有效性。

关键词: “货到人”拣选系统, 多机器人任务分配, 机器人调度, 鲁棒双层规划, 遗传算法