To address the problem of cumulative error divergence in LiDAR simultaneous localization and mapping(SLAM) and inertial navigation for indoor mobile robots in geometrically degenerate environments such as long corridors and glass curtain walls

an auxiliary localization and drift correction method that integrates AprilTag visual markers is proposed.The proposed method constructs a multi-level coordinate transformation chain model involving the map

odometry and camera

and estimates the observed pose of wall-mounted markers using a perspective-n-point(PnP) algorithm.A reprojection error evaluation mechanism is introduced to eliminate abnormal observations

and the pose difference in the Lie algebra space is used to update the drift compensation matrix from the odometry to the global frame in real time.Experimental results demonstrate that in complex scenarios with dynamic interference and transparent media

the proposed method can effectively suppress localization drift.The trajectory root mean square error (RMSE) is reduced by 92.4% and 64.7% compared with pure inertial navigation and LiDAR methods

respectively

which significantly improves localization accuracy and robustness in unstructured indoor environments.