Localization is a critical issue in autonomous robot navigation and path planning. In this paper, a novel localization method based on adaptive weighted fusion (AWF) is proposed. The method incorporates visual recognition of artificial landmarks and employs different adaptive fusion algorithms based on the pose and position of the mobile robot. Precisely, the method consists of three steps. First, a discrete motion model of the mobile robot is established. Second, the NDT-based laser rangefinder localization method is employed and visual odometry is replaced with artificial landmark recognition, with an analysis of the error model of the IMU and odometer. Finally, Adaptive Weighted Fusion (AWF) algorithm is utilized to fuse the sensor information mentioned above and correct the position and direction of the mobile robot. Experimental results demonstrate that the proposed AWF localization method can significantly improve the localization accuracy of the mobile robot. This research direction is promising and is expected to have broader applications in practical settings.