Indoor Localization Based on Fusion of AprilTag and Adaptive Monte Carlo

Abstract

Ordinary wheeled mobile robots use odometry and lidar to achieve indoor localization, but the localization accuracy of this method will be affected by odometry error. Considering that the AprilTag has good real-time performance and high local positioning accuracy, this paper proposes an Adaptive Monte Carlo Localization (AMCL) algorithm integrated with AprilTag. Take AMCL as the global localization framework, when the camera recognizes the tag, the absolute poses obtained by AprilTag are used to correct the error of the odometry model, improve global localization accuracy through local pose correction. Finally, a mobile robot is used for the actual field test, and the results show that the localization accuracy and stability are significantly improved after fusing AprilTag compared with the AMCL.