Research On AGV Dynamic Path Planning Based On Digital Twin

As an important automated logistics equipment,automatic guided vehicles AGVs have the advantages of high handling efficiency and flexibility,and are the main body for performing handling operations in modern storage systems.Reasonable AGV path planning is the key to guarantee the operational efficiency and stability of the system.The use of digital twin technology,real-time monitoring and conflict prediction of AGV operation state are carried out by establishing virtual model of physical entity to realize dynamic path planning,which has important practical guiding significance for improving AGV operation stability and system operation efficiency.This paper takes the goods-to-person picking system of enterprise S as the research object and study the dynamic path planning problem of AGV in the system,using digital twin technology and deep reinforcement learning algorithm to develop corresponding solutions.The main research elements are as follows.(1)Research on the dynamic path planning process and virtual model building of AGV based on digital twin.Based on the five-dimensional model of the digital twin and the enterprise reality,this paper construct a digital twin framework for the goods-to-person picking system,clarify the functions performed by each part of the framework in the path planning process,and design a dynamic path planning process for AGV based on the digital twin;analyze the elemental attributes and map environment of the physical entities in the system and establish a virtual model,and realize the monitoring of the AGV operation process by the virtual model with the help of the mapping of the virtual model to the physical entities.(2)Improvement of dynamic path planning algorithm based on deep reinforcement learning.To address the problems of slow convergence and low learning efficiency of traditional algorithms,an adaptive exploration strategy is designed by constructing the mapping relationship between reward and greedy strategy;at the same time,considering the dynamic scenario of a large number of AGV running simultaneously,a heuristic reward function consisting of gradient reward,gravitational reward and repulsive reward is designed for dynamic obstacles to improve the exploration efficiency and learning efficiency of the algorithm and realize dynamic path planning of AGV.(3)Research on conflict prediction and obstacle avoidance mechanism.Based on the digital twin virtual model to monitor the AGV operation process,monitor the nodes that the AGV is about to drive into,discover the potential conflict in time and judge whether the conflict occurs,discern the conflict type after confirming the conflict occurs,and select the corresponding obstacle avoidance strategy according to different conflict types to cope with the potential conflict during the AGV operation process.(4)Verify the effectiveness of the dynamic path planning scheme of AGV based on digital twin.The virtual models of personnel and equipment,map environment,path planning logic and AGV handling logic are built using Plant Simulation software to realize two-dimensional and three-dimensional virtual model runs,and the operational data are exported through data tables to realize visual monitoring of the operational data.Finally,the effectiveness of the scheme is verified by the virtual model operation results,which is of great practical significance to improve the stability of AGV operation and reduce the occurrence of conflicts.