Research On Localization And Mapping Technology Of Mobile Robot Based On Multi-source Information Fusion

With the in-depth development of the mobile robot industry,the tasks undertaken by mobile robots are becoming more and more diverse,and the environments they encounter are becoming more and more complex.How to achieve accurate positioning and correct mapping of mobile robots in complex environments is a matter of moving The prerequisite for robots to achieve autonomous movement and perform tasks.At present,the information obtained only by a single sensor cannot meet the working needs of mobile robots in different environments.In order to solve this problem,this thesis mainly studies how to use multi-source information fusion to help mobile robots achieve stable positioning and mapping functions in different environments.It mainly includes the following parts:(1)Through the analysis of the principles and models of various sensors,the characteristics of the information obtained by various sensors are studied,and the effect of learning from each other’s strengths and complementing each other’s weaknesses is achieved by using the information of different characteristics.In this thesis,the angular velocity information obtained by IMU is used to correct the laser lidar point cloud information of the rotating structure,Use factor graph models to fuse all the information.(2)In the traditional method,the odometry constructed by using multi-line lidar to extract edge features and surface features performs well in scenes with rich feature information,such as urban roads and ordinary indoor environments,but it is not easy to extract edge features in rural roads,open squares,etc.The error is larger in the environment of surface features.Aiming at the limitations of lidar for motion estimation in some scenes,this thesis proposes a lidar inertial odometry that uses IMU information and laser point cloud information to estimate inter-frame motion,which achieves more accurate estimation of inter-frame motion.(3)The error of the Li DAR inertial odometry will accumulate with the growth of the system running time.When the accumulated error is too large,it will lead to the failure of the positioning and mapping system.In order to reduce the cumulative error of the lidar inertial odometry,the factor graph model is used.Introduce loopback detection information and GPS information to optimize the positioning and mapping system,reduce the cumulative error of the lidar inertial odometry through loopback detection,and introduce GPS signals to provide global coordinates,which can make the system have the ability to run stably for a long time.(4)Design experiments in real scenes to verify the positioning accuracy and mapping effect of the multi-source information fusion positioning and mapping system proposed in this thesis,build a data acquisition platform in the real environment,and collect small indoor scenes,large indoor scenes,and outdoor scenes in the real environment.Open the data in the three environments of the large scene,realize the function of positioning and mapping in different actual scenes,and compare the positioning and mapping methods of the other two open sources,it proves that the multi-source information fusion positioning proposed in this thesis is effective.The mapping system has good positioning accuracy and mapping effect in different scenarios.