GNSS can provide high-precision positioning worldwide and is the preferred positioning method in autonomous driving and intelligent transportation etc. However, in complex urban environments, due to serious signal occlusion, the positioning performance of GNSS deteriorates sharply, and it even provides incorrect positioning information. To obtain robust navigation, GNSS is usually integrated with inertial measurement unit (IMU) to form a GNSS/INS integrated system. But in complex scenarios, the performance of GNSS/INS is closely related to IMU. For low-cost MEMS IMU, though it can facilitate GNSS positioning to a certain extent, it still cannot stably provide reliable positioning information. Visual sensors and IMUs can be combined to form a VINS system, which can obtain accurate local pose estimation. Therefore, adding visual information to MEMS IMU-based GNSS/INS systems can effectively suppress the divergence of MEMS IMU errors, and provide precise and reliable positioning services in GNSS-challenging environments.