Laser Inertial SLAM System With Reliable Plane Estimation And Keyframe

Simultaneous Localization And Mapping(SLAM)technology is one of the basic technologies of the positioning perception module of driverless technology,which provides information such as the carrier’s own position and attitude.This technology mainly uses sensors such as lidar and cameras to collect information about the vehicle environment,and matches with historical data,and finally obtains the position and posture of the vehicle through calculations.In order to achieve high precision localization and mapping in specific scenes,a laser inertial SLAM system with reliable plane estimation and key frame is designed in this study.The system can estimate the carrier pose with high precision in complex outdoor scenes.The main research contents are as follows:(1)It is proposed to progressively optimize the pose of adjacent lidar frames.First,the integrated pose of the inertial measurement unit is used as the initial value of the pose estimation。then,the matching pairs of feature points between adjacent radar frames are used to optimize the pose estimation,so as to avoid excessive errors in the integrated pose and affect the sliding window optimization results.(2)The concept of reliable plane is proposed to reduce the plane ambiguity in the denser local map point cloud.The plane feature matching pairs with the same feature points are correlated to obtain a reliable plane,and finally a more accurate plane residual constraint is generated.(3)A key frame mechanism is introduced,which uses the number of reliable planes to measure the richness of structural information in the local map point cloud,and then completes the key frame screening.This mechanism ensures that the local map can retain structural information to the greatest extent during the update process,and at the same time minimize redundancy.In this paper,qualitative analysis is made on the point cloud map built by the SLAM system designed by this research in the real scene,and quantitative analysis is made on the open data set MVSEC,and comparison is made with the current frontier location algorithm.The experimental results show that the SLAM system designed in this study can effectively improve the localization and mapping accuracy of the algorithm by optimizing the pose estimation of adjacent radar frames,introducing reliable plane estimation and key frame screening mechanism.