Abstract: “Part-to-picker” picking system adopts automated guided vehicle(AGV) to realize automatic picking operation. Shelves are transported by the AGV to the picking station, where they are used by the pickers to pick the goods. As the preparation for picking operations, order batching is a key factor affecting the number of AGV transporting and manual picking times. Optimizing the order division strategy is critical to improve the efficiency of the “part-to-picker” picking system. In the context of e-commerce intelligent warehouse, practical factors such as order demand for multiple goods, multi-shelf distributed storage of goods, unknown matching relationship between supply and demand of orders and shelves, etc., need to be comprehensively considered. Based on this, the objective of mathematical model construction is to minimize the sum of manual picking cost and AGV transporting cost. To solve the problem, an improved adaptive genetic algorithm is designed. The algorithm uses a heuristic strategy to generate the initial population, introduces crossover and mutation operators with adaptive transformation probability, and adds a local search process to enhance the ability of optimization. Finally, the validity of the model and algorithm is verified by experimental tests, and the superiority of the population initialization method is proved. According to the result, the sensitiveness analysis is used to give the reasonable allocation suggestion of turnover box quantity. This study can not only provide practical guidance for e-commerce enterprises to improve picking efficiency and reduce picking cost through order batch optimization, but also provide scientific basis for the practical application of “part-to-picker” picking system.

Key words: “part-to-picker” picking, order batching, AGV automatic guided vehicle, improved adaptive genetic algorithm