Research On SLAM Algorithm Of Outoormobile Robot Based On 3D Lidar

Nowadays,With the development of science and technology,mobile robots are widely used in fields such as unmanned driving,special scene inspection,and unmanned delivery.The simultaneous localization and mapping(SLAM)technology is a key prerequisite for the normal operation of mobile robots.However,the outdoor environment is complex and variable,such as uneven ground and various obstacles on the road,which can have an impact on the navigation of mobile robots and adjacent object recognition.In addition,during the outdoor operation of mobile robots,it is inevitable to encounter duplicate positions,which will also have an impact on establishing a globally consistent map and maintaining positioning accuracy for mobile robots.This paper considers that 3D Li DAR is not affected by outdoor light and has high-precision sensing function,so 3D Li DAR is used as the sensor for mobile robots.On the basis of the classic laser SLAM framework,this paper proposes a mobile robot laser SLAM algorithm that can operate stably and efficiently in complex outdoor environments.The main research content is as follows:Firstly,in order to address the complex outdoor road environment,the algorithm performs ground segmentation on the initial environmental point cloud scanned by the 3D Li DAR.The ground segmentation module encodes the initial point cloud into a representation based on a concentric region model,and allocates appropriate point cloud density between bins in a computationally uncomplicated manner.Then perform regional ground plane fitting to estimate the partial ground of each bin.Finally,ground likelihood estimation is introduced to reduce misestimation of the ground.Experiments have shown that adding regional ground segmentation can better cope with complex road environments,improve the accuracy of feature point matching,and enhance the robustness of the algorithm.Then,in order to solve the problem of duplicate position detection,this paper uses the original environment laser point cloud to construct a global descriptor-scan context for loop detection.Firstly,a KD tree is constructed using ring key vectors,and nearest neighbor search is performed to find multiple frames and their ring key values that may be similar to the current frame.Then calculate the similarity score and filter out similar frames with higher scores.Next,calculate the minimum offset and similarity score for each sector,and select the frame with the highest similarity score as the loop frame.Finally,the loop detection was successfully completed.After analyzing the experimental results,it was found that scan context can improve the efficiency of algorithm loop detection and reduce the error between the algorithm trajectory and the real trajectory.Finally,the SLAM algorithm in this paper was tested and analyzed using the mobile robot platform and the open-source KITTI dataset,respectively.At the same time,the laser SLAM algorithm in this paper was compared with the popular laser SLAM algorithm.The experimental results demonstrate that the algorithm proposed in this paper can improve the positioning accuracy of mobile robots in outdoor environments and establish a 3D laser point cloud map that is more in line with the actual environment.Compared with the popular laser SLAM algorithm,it has also been proven that the algorithm proposed in this paper can better adapt to outdoor environments and improve the efficiency of loop detection.