Research On Multi-AGV Path Planning In Large-scale Logistics Sorting System

With the rapid development of e-commerce industry,the fast growth of express delivery in China has brought great challenges to the logistics industry.In order to alleviate the increasing pressure of the logistics sorting center,logistics enterprises have expanded the site of the logistics sorting center and introduced the Automated Guided Vehicle(AGV)to sort objects automatically and intelligently.Therefore,the issue that how to improve the sorting efficiency of AGV system in large-scale logistics sorting center has attracted the attention and thinking of enterprises.This paper studies the multi-AGV path planning of logistics sorting center in large-scale scenario and mainly completes the following work:Firstly,this paper studies and analyzes the significance of multi-AGV path planning,and then investigates the related algorithms of AGV path planning.Combined with the actual situation,this paper uses the mainstream rasterization modeling method to model the site of the logistics sorting center,and introduces the model of multi-AGV path planning,classic A* algorithm and Conflict-based Search algorithm.In the large-scale logistics sorting center environment,this paper selects the sum of the path costs of all AGVs as the optimization target,and proposes the offline multi-AGV path planning algorithm named DC-CBS algorithm.The DC-CBS algorithm combines the idea of the divide-and-conquer algorithm with the CBS algorithm,making full use of the idea of dividing the sub-problem by the divide-and-conquer algorithm and the excellent performance of the CBS algorithm in a small-scale environment.The DC-CBS algorithm divides the large-scale scene map into several identical small-scale blocks and maps them into upper-level maps.Using the local optimal strategy,the original multiAGV path planning problem is decomposed into multiple similarly ones with less computation.Finally,the solution of the sub-problem is combined into the final solution of the original problem.The results indicates that the total cost of the path solution planned by DC-CBS is almost equal to the optimal solution while the running time of DC-CBS algorithm is less than that of CBS algorithm in a larger and smaller scale.Finally,this paper proposes an online path planning algorithm with lock grid,which solves the problem that the actual environment has higher requirements on the first response time.The lock grid is a safe distance set by the algorithm.Each path plan shouldrender all AGVs conflict-free.The path planning algorithm with lock grid also simplifies the conflict resolution and speeds up the algorithm.The results shows that the first response time of the path planning algorithm with lock grid is shorter than that of CBS algorithm,and the time of the first response is very close for different AGV quantities.In addition,this paper simulates the actual scene map of the logistics sorting center.The simulation interface not only shows the layout of the logistics sorting center and the distribution of the AGV,but also dynamically display the moving process of the AGV.Considering the fact that the probability of obstacles affecting the operation of AGV is high in the actual environment,the simulation experiment adds the function of dynamically adding obstacles at the interface,enabling AGV to handle static and dynamic obstacles intelligently.