Research On Two-level Path Planning And Simulation Of AGV System In Manufacturing Workshop

With the continuous improvement of consumption level,there will be more and more challenges for Job-shop.The production efficiency of Job-shop needs to be improved,and intelligence is the best solution.This is also in line with the background of industry 4.0 and is an important part of the "made in China 2025".As an important part of flexible manufacturing unit,AGV is an indispensable part of intelligent factory and intelligent manufacturing.In this paper,AGV technology,AGV system and AGV path planning methods at home and abroad are investigated.It is found that in the research of AGV,the path planning of AGV is always a difficult and important point.Most of the current path planning algorithms are carried out in the abstract topology map.This is very important for the theoretical research of AGV path planning algorithm,but the topological map is quite different from the actual production workshop.With the continuous update of AGV hardware platform,AGV itself has the function of real-time location and mapping.Based on the warehouse of a manufacturing workshop in Shenzhen,combined with the new AGV,this paper puts forward the double-layer path planning framework of AGV system.In the decision-making layer of AGV system,the first layer path planning of AGV is carried out based on the abstract topology map to determine the nodes that AGV passes through in turn when it performs tasks;in the AGV layer of AGV system,dynamic path planning is carried out in a small range based on the real-time built occupation grid map to solve the AGV conflict.The advantages of traditional algorithm theory and new AGV are integrated.Based on this framework,the main contents of this paper are as follows:(1)A two-level path planning algorithm is proposed for multi AGV path planning.The first layer of bi level path planning is an improved A-star path planning algorithm.By analyzing the traffic areas of AGV in different states,the path planning of AGV is carried out,and the possibility of multiple AGV conflicts is reduced while planning for a single AGV.The second layer of bilevel path planning is a RRT dynamic path planning algorithm combined with priority strategy.Point cloud grid map based on AGV real-time construction can effectively avoid path conflict between AGVs and improve the efficiency of AGV system by searching local grid map to dynamically solve local path when AGV conflicts or encounters obstacles;(2)According to the two-level path planning algorithm,the AGV system is constructed.The system mainly consists of upper computer management control layer,AGV layer and communication layer.It is a semi hybrid system,which combines the advantages of distributed system and centralized system.The upper computer management and control layer controls AGV in a centralized way,and makes the first level path planning for each AGV in the global topology map.Distributed management is implemented in AGV layer,and each AGV plans its own path in local grid map.The communication layer connects the two layers to ensure the efficient and accurate operation of AGV system;(3)The two-level path planning algorithm is verified by simulation.At first,the improved A-star algorithm is verified,and compared with A-star algorithm and Dijkstra algorithm,the improved A-star algorithm is proved to be more efficient in searching,and also significantly reduce the possibility of multiple AGV conflicts.Then,comparing the two-level algorithm with other multi AGV algorithms,the production path of the two-level path planning algorithm is more optimized,and the time-consuming is far shorter than other algorithms;(4)According to the double-layer path planning algorithm and the kernel of openTCS,the AGV system is developed.The display and control interface that interacts with users is developed.The whole operation process of the system is described,and the single AGV and multi AGV are scheduled by using the AGV system,which proves its functionality.