移动机器人多传感器融合建图与定位算法研究

iming at the problem that the accuracy and robustness of the existing Lidar point cloud mapping and visual mapping are low in complex environment, a multi-sensor fusion mapping algorithm is proposed in this paper. The Lidar point cloud information and visual information are fused in the process of mapping, and the Lidar point cloud-visual fusion map based on key frame is constructed. Two kinds of information are used at the same time to detect the loop and estimate the pose from coarse to fine, and the error of fusion map is reduced by the pose map optimization. Based on the globally consistent fusion map, accurate and robust global and local localization can be achieved. Based on the actual data collected, the test results show that the proposed method has better mapping and positioning results than only using Lidar point cloud or vision.