Research on SLAM of Corridor Environment Based on Multi-Sensor

In view of the problems of large positioning deviation and map offset in the use of laser and vision sensors to construct maps in the corridor environment, the current stage of multi-sensor fusion SLAM algorithm is researched. Improving a SLAM algorithm based on weighted observation fusion EKF, fusing lidar, depth camera and IMU sensor information, and adding a closed-loop detection verification mechanism at the back of the SLAM algorithm. In order to verify the effectiveness of the algorithm, 16 feature points are selected to perform error analysis. The average map update time is reduced by 0.29s, the average relative error is reduced by 1.277%, and the maximum relative error is reduced from 3.130% to 0.673%.