Key points are not available for this paper at this time.
Integration of global navigation satellite systems (GNSS) with other sensors, such as inertial measurement units (IMU) and visual sensors, has been widely used to improve the positioning accuracy and availability of the vehicles for the Internet of Things (IoT) applications in smart cities. The traditional extended Kalman filter (EKF)-based fusion scheme, with the assumption of fixed measurements of different sensors and inaccurate GNSS quality assessment, is vulnerable to non-line-of-sight (NLOS) and multipath contaminated GNSS, as well as low-quality vision measurement. In order to tackle this issue, we have proposed an adaptive weighting strategy for GNSS/IMU/Vision integration. On the basis of dual-check GNSS assessment, we adjust the weights of the vision and GNSS measurements adaptively based on the chi-square test statistic. The field tests have demonstrated that the proposed algorithm achieves horizontal positioning root mean-square errors (RMSEs) of 11.92 and 3.61 m in deep and mild urban environments. The accuracy has improvements of 78.57% and 43.9% over traditional EKF-based GNSS/IMU fusion, and 21.53% and 23.49% over compared EKF-based GNSS/IMU/Vision fusion, respectively.
Sun et al. (Mon,) studied this question.