Path Planning For Multi-AGV Systems

Automated Guided Vehicle(AGV)is not only used in large enterprises,but also favored by more and more small enterprises.Path planning technology is the core of AGVs system.This paper studies the path planning problem of magnetic navigation AGV.The goal is to design a path planning scheme that can guarantee the efficient and collision-free operation of multiple AGVs.This paper studies the multi-AGV path planning scheme based on two-stage control strategy.Taking the implementation process of the scheme as the main line,three problems are studied:the K-shortest loopless path algorithm for generating offline path libraries,the multi-group genetic algorithm combining offline path libraries with genetic algorithms,and the solution framework for AGV real-time conflicts.The main work of this paper consists of three parts:(1)An algorithm for generating a K-shortest loopless path is proposed.The algorithm is based on the classical Dijkstra algorithm.During the process of the algorithm,the single-source K-shortest loopless path is calculated by dynamically adjusting the threshold of the number of paths.The algorithm is implemented in the paper and compared with other solving algorithms,the results show that the algorithm is intuitive,simple to implement,and has good time complexity,which can meet the requirements of constructing offline path library in multi-AGV scheduling.(2)A path planning scheme based on two-stage solving strategy is proposed to solve the path planning problem of multiple AGVs:the path library is generated by the acyclic K-shortest path algorithm in the offline phase.In the online phase,a multi-group genetic algorithm is used,which consisted of a primary population and a secondary population.It uses the offline path library in evolution to optimize the path,and judge whether the path scheme is good enough by considers the running time,the number of turns and stop times,and the delay time of the task.(3)Taking multi-AGV real-time scheduling as the research object,The conflict handling method is proposed to solve the real-time conflict problem caused by the new task joining,road block blocking,and intersection blocking.