An Adaptive and Robust Strategy for GPS/IMU/VO Integrated Navigation

The multi-sensor integrated navigation is critical for improving the continuity and accuracy of positioning for intelligent transport system (ITS) applications. Global Positioning System (GPS), inertial measurement unit (IMU) and visual odometry (VO) are commonly used sensors due to their complementarity. In complex urban environments, multipath interference (MI), non-line of sight (NLOS) and insufficient satellites degrade the positioning performance. Therefore, VO is required to correct the accumulated IMU errors and unreliable GPS positioning. Filter algorithms are commonly used in multi-sensor fusion. In the traditional Kalman filter, the measurement noise matrix and the fusion strategy of multi-sensors are usually fixed, which cannot adapt to the surrounding environment and the measurement quality of different sensors. To address this issue, we propose an adaptive and robust strategy for GPS/IMU/VO integrated navigation system. For the construction of the measurement vector, a dual GPS quality check method is proposed to realize adaptive switch of the filter. We also construct the robust factor weight matrix, so as to reduce the interference of gross errors on filter estimation. The field test in urban environment shows that the horizontal and 3D positioning accuracy of the proposed algorithm is 5.45 m and 5.83 m. Compared with the traditional GPS/IMU and GPS/IMU/VO fusion algorithm without adaptive and robust strategies, the 3D positioning accuracy has improvement of 69.49% and 44.64%, respectively.