Research On Key Technologies Of High-Speed Sorting Robot Scheduling For Small Parcels

In warehousing operation costs,sorting operation accounts for an increasing proportion,about 55%.At present,the small parcel sorting system in use has expensive equipment,large area and fixed sorting scale.Therefore,a high-speed sorting system for small parcels that can dynamically adjust the scale of the system based on demand provides a new solution for logistics enterprises.It can flexibly use different type and number of robots and dynamically arrange worksites of different sizes to adapt to the fluctuations in the number of orders.This system can reduce equipment loss and energy consumption,which meets the requirements of carbon neutrality policy.This paper studies the layout planning problem before the construction and the robot scheduling problem in the sorting operation of this new sorting system.The specific work of this paper includes:(1)Based on the two-stage queuing process,a layout model for sorting robot number and site requirements before the construction of the sorting system that adapts to the dynamic workload is constructed.This model,using the parcel sorting throughput time as performance index,is verified by an example.It is convenient to use this model to obtain the optimal rectangular sorting area size and sorting robots number under a specific workload;(2)Based on the minimum sorting cost and the shortest sorting time,a three-dimensional scheduling model of parcel-carry robots,box-carry robots and sorting ports(turnover box placement positions in sorting area)is constructed.According to the characteristics of the scheduling model,a two-stage three-layer feasible solution code is proposed,and a hybrid scatter search algorithm combining exchange operator(exchange sorting tasks performed by sorting robots),greedy operator(reduce the number of working robots)and Tabu search algorithm is designed to solve the model;(3)The MAPF benchmark dataset proposed by Nathan Sturtevant is improved to validate the model and algorithm described in(2).Compared with the genetic algorithm and the standard scatter search algorithm,the accuracy of the proposed algorithm is improved by 10.3%and 6.5%respectively,and the convergence speed is improved by 15.5%compared with the standard scatter algorithm.