This paper focuses on the problem of moving object detection (MOD) in robot visual simultaneous localization and mapping (vSLAM) system. A MOD algorithm is designed using the spatial geometric constraint of the stationary landmarks in the environment. Based on the MOD algorithm, the moving objects can be discriminated from the stationary landmarks. The proposed MOD algorithm is independent of the state estimator and capable of dealing with the kidnapping problem in SLAM automatically. Meanwhile, the method of speeded-up robust feature (SURF) is employed in the algorithm to provide a robust detection for image features as well as a better description of landmarks in the map of a visual SLAM system. Experiments are carried out on hand-held camera sensors to verify the performances of the proposed algorithms for SLAM tasks in the indoor environments. The results show that the integration of MOD and SURF is efficient to improve the robustness of robot SLAM system.