Abstract: In this paper, a method of robot pose correction based on multi-source information fusion using extended Kalman filter (EKF)and point-line iterative closest point(PL-ICP)point cloud matching algorithm is proposed to solve the problems of large cumulative error and low accuracy of single pose estimation method for robots. In order to reduce the cumulative error caused by the wheel odometer, Mahony algorithm is used to calculate the attitude of the gyroscope and accelerometer. The preliminary estimation of robot pose is obtained by fusing the wheel odometer based on the extended Kalman filter. In order to reduce the influence of wheel deformation and slip on the pose of the robot, using the PL-ICP point cloud matching algorithm to construct a single-line laser odometer to estimate the robot pose again; In order to improve the accuracy of pose estimation, based on the total mean square error of the two poses and the pose error at the time before and after, an adaptive correction algorithm for the cumulative error is constructed. The method in this paper obtains global optimal weight factor and local dynamic weight factor to realize the adaptive adjustment of the cumulative error correction factor by analyzing the total mean square error of the two poses and the pose error at the time before and after, which can obtain more accurate pose estimation of the robot. Experimental results show that this method can correct the cumulative error of the robot's pose, and significantly improve the accuracy of robot pose estimation.

Key words: laser odometer, EKF, robot pose estimation, PL-ICP, multi-source information fusion