Abstract: Map updating algorithms in outdoor semi-static environments generally suffer from slow speeds and inability to run in real time. Therefore, a map updating method based on fast session alignment is proposed to improve the session alignment speed and ensure consistency between the map and the environment. Firstly, Gaussian curvature is introduced to sparsify point clouds and reduce their size, followed by voxel based point cloud registration, and the obtained constraints are used to construct factor maps. Then, historical constraints are introduced to implement factor graph repair for local sessions, solving the problem of factor graph construction failure caused by loop detection and point cloud registration failure.Finally, session alignment is achieved through factor graph optimization, and map updating is performed based on the aligned sessions. The method is evaluated on MulRan dataset and LT-ParkingLot dataset, and its session alignment frequency reaches 13 Hz, which is 80% higher than the original method, and sufficient for real-time session alignment in classic semi-static scenarios such as factories. The effectiveness of the proposed method in map updating is demonstrated through experiments.