Research On Integrated Navigation Technology Based On Inertial/Vision/Lidar

The combined navigation technology based on inertial/vision/lidar is a positioning method that uses the input of three sensors of IMU,camera and lidar as the navigation information source.It has the characteristics of less dependence on the external environment and strong adaptability.Suitable as a navigation and positioning system for mobile robots.Therefore,this paper focuses on the following aspects based on the fusion of data between sensors in inertial/vision/lidar combined navigation technology.Firstly,the coordinate systems commonly used in inertial/vision/lidar combined navigation technology and their conversion relationships are introduced,the inertial/vision/lidar sensor model is analyzed,and the pose estimation method of a single sensor in navigation technology is studied.Then,for the fusion between inertia and vision,an inertial vision mileage calculation method based on IMU pre-integration is given.This algorithm first performs feature extraction and tracking of visual data,and at the same time pre-integrates the IMU data,and then The system is jointly initialized,and then the position and attitude of the inertial visual odometer are estimated by means of nonlinear optimization.Finally,the effectiveness of the algorithm is verified through experiments.Next,for the fusion of inertial vision odometer and lidar,an inertial vision radar pose estimation algorithm based on line feature extraction is given.The algorithm first performs line dividing,distortion removal and feature extraction on radar data Preprocessing operation,then use the ICP algorithm to jointly estimate the pose of the inertial vision odometer and lidar,and finally verify the effectiveness of the algorithm through experiments.Finally,the ROS platform is used to design the simulation experiment system of the inertial/vision/lidar combined navigation algorithm,and the system is simulated using the KITTI data set.In order to verify the adaptability of the algorithm to the environment,the experiments were conducted in two different scenarios.The results show that the algorithm has the highest accuracy in scenarios where both vision and lidar are suitable for navigation;in scenarios where neither is suitable for navigation,The algorithm has the lowest accuracy;and,under the same scene,the accuracy of the combined inertial/vision/lidar navigation algorithm is better than VINS.