AGV Localization Based on Odometry and LiDAR

Localization is the key to AGV autonomous navigation. The motion model and the observation model of the AGV are established based on odometry and LiDAR respectively. It is proposed that the estimated position of the AGV is obtained by the odometry, then the accurate location is quickly obtained by matching observed information. The cumulative error of the motion model and the measurement noise of the observation model are considered, the latter is subjected to probability filtering to correct the odometry in the SLAM process. An improved Monte Carlo localization method is proposed, which increases the screening pass rate of effective particles in the initial localization process and improves the efficiency and accuracy of localization.