Research On Multi-sensor Fusion Based Indoor Localization System Of Mobile Robot

In recent years,the development of intelligent mobile robots has spawned various concepts and industrial applications.The application fields and demand ranges of intelligent mobile robots in traditional manufacturing have been expanding.Localization is the core function of all mobile robots.For complex indoor working environments,it is of great significance to carry out research on mobile robot localization technology based on multi-sensor fusion.In this article,we established a multi-sensor fusion robot indoor localization system to improve the accuracy and robustness of localization.Mechanical hardware,and technologies including remote communication,and map construction were adopted and combined.The main contents of the thesis include:Firstly,for the indoor working environment with weak textures and weak structural features,the hardware of the indoor localization system for mobile robots is selected,and an experimental platform for mobile robot localization,which include wheel encoder,IMU,vision camera,and LIDAR,is designed and built.The communication network module of the robot operating system was used to realize remote control and communication of mobile robots.Gazebo simulation platform was further built to establish a simulation test system.Secondly,we analyzed the advantages and disadvantages of the wheel encoder and IMU.We further established a track estimation algorithm with the information of wheel odometer and IMU used to obtain the fused odometer information,which improve the localization accuracy of the robot.Indoor Localization error was further analyze based on the improved algorithm.Finally,through comparative experiments,it was verified that the proposed fusion localization algorithm could improve the indoor localization accuracy of mobile robots.Thirdly,a localization and mapping system based on the combined odometer information and vision camera was implemented.Aiming at the instability of the pure visual localization system in a weakly textured environment,a visual feature point tracking algorithm and a post-fusion odometer were used to solve the camera pose,and a simultaneous localization and mapping system was constructed.Through comparative experiments,the proposed algorithm could improve the robustness of mobile robot system.We further ran the algorithm in a real scene to build a visual map.Finally,in order to solve the instability of pure laser sensor localization in a weakly structured feature environment,a prediction model was established with the above fused odometer,and a measurement model was established with the camera pose and LIDAR to achieve multi-sensor fusion localization algorithm based on the particle filter system.We further established laser map.We also put forward the motion path matching algorithm using the results of laser map and visual map to achieve multi-source map matching.Based on the results of this multi-source map,we improved the accuracy of localization.We further ran the algorithm in simulation and real scenarios,and the results show that the algorithm proposed in this paper had high localization accuracy and robustness.