Research On Simultaneous Localization And Mapping Algorithm Based On Regional Map Reconstruction In Polar Coordinates

At present,the research on the simultaneous localization and mapping(SLAM)algorithm based on lidar is mainly focused on the Cartesian coordinate system.This ignores the lidar discrete and even sampling mode of polar coordinates.In view of this characteristic,this paper proposes a SLAM algorithm in polar coordinates.First of all,for the raw data of lidar,a map reconstruction algorithm based on the regionalized Gaussian process in polar coordinates is designed in this paper.In the reconstruction process,the polar angular position of the reconstruction result is fixed,which is used to directly determine the pairing point relationship and eliminate the mismatching problem caused by discrete sampling.The purpose of equal regionalization is to treat the data in different subregions differently,to ensure that the Gaussian process completes the map reconstruction along the appropriate direction.Then,based on the map reconstruction results,a corresponding laser registration algorithm is designed in polar coordinates.The registration algorithm performs rotation estimation and translation estimation step by step,and obtains the pose relationship between the sensor and the world map at the current moment.Among them,the rotation estimation uses the rotation characteristic represented by the polar angle to quickly find the optimal rotation angle;the translation estimation uses the weighted least squares method to minimize the difference between the polar diameters and obtain the translation distance.Aiming at the map reconstruction result after registration,according to its regionalization characteristics,a map fusion strategy based on the subregion layer and point cloud layer was proposed.The variance information carried in the reconstruction result was used to perform a world map recursive least square update.The results of multiple controlled experiments show that the algorithm proposed in this paper can perform simultaneous localization and mapping tasks well in simulation tests and real experiments,indoor and outdoor environments,and sparse and dense data sets.At the same time,it is superior to current mainstream two-dimensional SLAM algorithms in both accuracy and efficiency.