Improved algorithm of cartographer based on laser odometer

Abstract

In the front-end matching process of the Cartographer algorithm, the accuracy of matching between the point cloud and submap relies on the initial values provided by the pose fusion algorithm. However, the original algorithm's pose fusion algorithm has low accuracy. To address this issue, this paper proposes an improved Cartographer algorithm based on a laser odometer. The improved algorithm utilizes NDT registration to obtain the pose transformation between frames. Additionally, a pre-integration of the IMU between the front and back frames is performed for joint optimization, allowing for the acquisition of a more accurate pose. This enhanced accuracy contributes to improving the matching of high point clouds with the submap. To analyze the efficacy of the improved algorithm, comparisons were made with the original Cartographer algorithm by analyzing the map construction effect and conducting positioning accuracy tests using datasets. The experiments confirmed that the improved algorithm is both feasible and effective in enhancing the map construction effect and pose accuracy.