| With the continuous advancement of science and technology,people's living standards have been continuously improved,and the demand for high quality of life has also increased.The emergence of home mobile robots has just filled people's needs.With the continuous development of artificial intelligence technology,the "intelligence" of home robots is also increasing.The improvement of SLAM technology for home mobile robots also represents an increase in artificial intelligence technology.In this thesis,the accuracy of the robot positioning,accuracy of construction,and real-time performance are improved through the improvement of the Graph SLAM algorithm.The main research content of this thesis is as follows:First of all,from the significance of researching home mobile robots,the importance of using home mobile robots to improve the quality of life of modern humans is described.Secondly,the domestic and foreign research status of the problem of real-time localization and mapping(SLAM)of mobile robots,as well as the common open source algorithms are introduced and compared in detail.Then the software operating system ROS used in this topic and the characteristics of the mobile robot's chassis Turtlebot2 are briefly introduced.The sensors used in this project,such as the construction of the odometry and the IMU's motion measurement model,and the laser radar measurement principle are described.Through the comparison of the map description methods,it is determined that this thesis takes the probability of the grid map as the map mode,and uses the form of the local submap to build the map.In view of the poor real-time performance of open-source algorithm Karto SLAM and the instability of loopback detection,this paper proposes two improvements.Firstly,the UKF filtering algorithm is used to fuse the odometer and IMU measurement data to improve the accuracy of positioning.Then,the branch-and-bound method of depth-first search is used to accelerate the pruning of the back end of the graph optimization to improve the real-time performance.Finally,the use of ceres optimization to complete the scan match,to further improve the positioning accuracy.This thesis summarizes the full text in the last chapter and looks forward to further research in the future. |
