Research On SLAM System Based On Fusion Of IMU And Variable-Height Lidar

Laser Simultaneous Localization and Mapping(SLAM)mainly uses lidar sensors for pose estimation and environmental map construction.It is widely used in field such as autonomous driving,space exploration,and mobile robotics.The related theory of SLAM based on lidar sensors is relatively mature,but there are also problems such as the restrictions on the price of lidar sensors,single-line lidar scanning point cloud data is less,the difficulty of accurately matching features in large rotating scenes,and the poor adaptability to the ground environment with slope.The main research contents of this paper are mobile robots based on the extended two-dimensional laser lidar scanning environment,extracting line and plane features to improve feature matching efficiency and accuracy,integrating inertial measurement unit(IMU)data to solve the problem of excessive robot rotation angle,and using laser point cloud perform pose compensation to improve map accuracy.In this paper,we propose a variable-height lidar odometry method to enhance the accuracy of the mobile robot motion estimation for an indoor environment.The lidar device adopted a novel variable-height lidar,the extraction of line segment features and plane features from the laser point cloud is proposed and an effective line segment and plane feature matching algorithm is designed to combine line segment features and plane features.Improved the accuracy and robustness of the laser SLAM system in a structured indoor environment.The turning angle of the mobile robot during the movement is too large,which causes the mismatch of the collected point cloud features.In this paper,variable-height lidar expands environmental information,based on the line and plane matching method,and pre-integrated information of IMU sensor is integrated.A SLAM algorithm that combines the measurements of various sensors in a tightly coupled manner is proposed,the linear optimization method solves the optimal state estimation by maximizing the posterior probability.In order to solve the map constructed cannot reflect the real environment at the mobile robot's undulating ground motion.Firstly,the laser point cloud used to construct the map are calibrated,then compress the calibrated laser point cloud into a two-dimensional laser,and finally construct a two-dimensional occupancy grid map by calibrated and compressed the laser.Based on the aforementioned algorithm,an experimental platform based on Pioneer 3-DX,variable-height lidar,and Xsens MTI-G-700 is built.The mobile robot pose estimation and environment map construction algorithm proposed in the paper is structured.A comparative experiment was performed in the indoor environment,and it was verified through experiments that the method discussed in this paper is superior to two-dimensional laser SLAM and traditional wheeled odometry calculation methods in positioning accuracy,map construction accuracy,and algorithm robustness.