Abstract:

In recent years, with the rapid development of intelligent driving technology, the requirements of vehicles for environmental perception, positioning, decision-making, control and other systems are increasing. As a prerequisite and basis for other systems, real-time and accurate positioning of a vehicle has important research values. The Global Navigation Satellite System (GNSS) can provide absolute positioning results without accumulative errors, but it has weak resistance to interference and low output frequency. Inertial Navigation System (INS), on the other hand, has a high sampling frequency and good real-time performance, but cannot be used alone for long periods of time due to the accumulative errors. As the combination of the two can effectively complement each other, the GNSS/INS integrated positioning algorithm is widely used as a mainstream positioning method in the field of vehicle navigation and positioning.

However, as the main working condition of a vehicle, an urban environment is complex, causing challenges for vehicle positioning systems. Tall buildings, trees and overpasses in an urban environment can block satellite signals, resulting in a significant reduction in GNSS positioning accuracy. When using the Kalman filter to fuse GNSS and INS measurements, a vehicle may receive anomalous GNSS positioning results, which can lead to reduced performance of the integrated vehicle positioning system and cause driving safety problems if the filter parameters are not adjusted in a timely and accurate way.

Aiming at the problem of a reduced positioning accuracy of the integrated vehicle positioning system due to GNSS signal anomalies in urban environments, this paper proposes a fuzzy adaptive Kalman filter-based integrated vehicle positioning algorithm. The algorithm makes full use of the auxiliary information provided by GNSS (the number of visible satellites and position dilution of precision), builds a fuzzy inference system to monitor them and outputs the adjustment coefficient of measurement noise. The improved Sage-Husa algorithm is then used to further estimate the measurement noise based on the filter innovation. Combining the above methods, the covariance of the measurement noise in the filter parameters is adjusted timely and accurately by fuzzy adaptive estimation. Using the vehicle position error, velocity error, orientation error and the error in the bias of the accelerometer and gyroscope in three axis as state vectors, the Error State Kalman Filter (ESKF) is used to fuse the GNSS and INS measurements by replacing the fixed filter parameters of the standard ESKF with the measurement noise covariance obtained by the fuzzy adaptive estimation algorithm described above. At the same time, with reference to the kinematic characteristics of the ground vehicle, the kinematic constraints of the vehicle are constructed to further correct the errors of the GNSS/INS integrated positioning system by means of measurement updates.

The performance of the proposed algorithm is verified by computer simulation and real vehicle experiments. The results show that the proposed fuzzy adaptive Kalman filter-based integrated vehicle positioning algorithm can adjust the measurement noise covariance in a timely and accurate way and correct the positioning error by vehicle kinematic constraints, and it can still provide accurate positioning results in areas where GNSS signals are anomalous. The positioning accuracy of the proposed algorithm is significantly improved compared to the standard ESKF algorithm.