Research And Implementation Of AGV's Multi-target Navigation System Based On ROS

As one of the core elements of modern intelligent logistics,AGV(Automated Guided Vehicle)plays an important role in the material handling and assembling of industry in manufacturing,tobacco,express delivery,medicine and electric trades.At present,a wide range of independent platforms of AGV increase the difficulty of development,and its single target navigation system can't meet the need of high flexibility of modern intelligent logistics.Therefore,the proposed AGV multi-target navigation system based on the highly versatile,compatible and multi-language programming supported ROS(Robot Operating System)and combined with the enterprise's actual needs is of important research value.ROS provides a complete,open-source,visual platform for robotic research.With its integrated tools and library programs,researchers can quickly simulate real world and begin experiments.Based on the advantages of ROS platform,the high-precision laser range finder is selected as the main sensor,and SLAM(Simultaneous Localization and Mapping)is applied for simultaneous positioning and map building,what's more,JSP(Java Server Pages)dynamic web technology is contained to design AGV's multi-station calling Web application,so an overall framework of the AGV multi-target navigation system based on ROS is established.First of all,by analyzing several mainstream SLAM algorithms,this thesis argues that the Rao-Blackwellized particle filter(Rao-Blackwellized Particle Filter,RBPF)provides an effective solution for SLAM.To solve the problem that RBPF-SLAM algorithm based on improved proposed distribution is difficult to construct an accurate priori map in dynamic environment,laser data preprocessing and improved particle sampling technique are introduced.In this way,the untrusted particles are eliminated by sampling from the previous generations particles,therefore,the anti-interference of the algorithm is improved.The experimental results verify the robustness of the proposed algorithm.Secondly,the multi-target path planning algorithms are studied,and to solve the problem that ant colony algorithm has low searching efficiency and easily falls into local optimum,an improved ant colony algorithm is proposed,by adaptively changing the ant population size,pheromone intensity and weight coefficient,introducing crossoperation,reverse operation from genetic algorithm at the same time,the search efficiency and the global search ability of the algorithm is enhanced.The experimental results show that the improved algorithm is less time consuming and has shorter path,when applying it to TSP(Traveling Salesman Problem),the global optimal traversal sequence of AGV multi-target navigation can be obtained quickly and efficiently.At last,the architecture of AGV multi-target navigation system under ROS platform is completed,and the AGV's multi-station calling Web application program is designed.In the real indoor environment,a series of tasks including the multi-station calling,indoor map construction,multi-target path planning,intelligent obstacle avoidance are completed by AGV.Experimental results show that the algorithms proposed in this thesis and the AGV multi-target navigation system are feasible and reliable.