Task Allocation Of Robotic Mobile Fulfillment System Based On Dynamic Partition Auction

Since warehousing is an important link in the manufacturing and supply chain,the improvement of warehouse efficiency is of great significance to ensure the normal operation of the manufacturing and supply chain.Currently,the Robotic Mobile Fulfillment System has become the preferred solution for the picking warehouses for multiple varieties and small batches of goods,and the improvement of their order picking efficiency is conducive to improving the responsiveness of warehouses.To improve warehouse operation efficiency,the goods picking process is transformed into an order task allocation problem.A task allocation method of Robotic Mobile Fulfillment System based on dynamic partitioning auction is proposed,to improve the task assignment efficiency,it utilizes auction mechanism and auction algorithm and dynamic partitioning for its task characteristics as large scale,dynamism,and complexity.This method not only improves the allocation efficiency of large-scale dynamic and complex tasks in order picking but also can be extended to the storage space allocation and replenishment of the Robotic Mobile Fulfillment System.The specific research works are as follows.Firstly,focusing on the task allocation problem of the Robotic Mobile Fulfillment System,the picking process in the system and its task allocation process are analyzed,and the MAS-based task allocation process model is thus established.Besides,based on the Agent modeling rules,task characteristics,and system scenario characteristics,the structural model and information exchange model of each Agent during the task allocation auction are particularly designed.Secondly,based on the constructed MAS single-round task assignment model,fuzzy logic and parallel auction are adopted for the auction mechanism to calculate the task bidding utility,and an improved sealed second price auction is designed to improve the efficiency of task allocation utility matrix calculation.The experimental results verify the effectiveness of the proposed improved auction algorithm in efficiency improvement.Then,for the traits of tasks in RMFS scenarios,the actual tasks are decomposed into several rounds of static task assignment,and matrix sparsification is proposed to process the utility matrix according to the static task characteristics.The utility matrix is conducted on dynamic partitioning,and the utility matrix within the partitioned area is solved by the improved auction algorithm to achieve a real-time allocation of largescale and dynamic,and complex tasks in RMFS.The performance of the dynamic partitioning method is verified by Python simulation experiments.Lastly,based on the Netlogo platform,an RMFS simulation system is built.The fuzzy logic-based dynamic partitioning task auction algorithm is applied in the simulation system,which verifies the effectiveness of the constructed RMFS simulation model and the proposed improved method and analyzes the superiority of the improved method compared with the traditional task assignment algorithm.