Vision-assisted GNSS/INS high precision positioning method based on adaptive maximum correntropy criterion in urban traffic environment

Accurate, continuous and reliable positioning services are the foundation of new generation information technologies such as intelligent transportation and smart cities. The global navigation satellite system (GNSS)/inertial navigation system (INS) integrated system has been widely used in the fields such as autonomous driving and mobile mapping. However, numerous obstacles in cities reflect and block the GNSS signals, generating multipath (MP) and non-line-of-sight (NLOS) errors, and even causing GNSS outages, significantly degrading the positioning performance of the GNSS/INS integrated system. In this contribution, a vision-assisted GNSS/INS high precision positioning method based on adaptive maximum correntropy criterion (MCC) and multiconstraint dynamic feature points elimination is proposed. First, visual positioning information is used to compensate the loss of GNSS observation information during GNSS outages, and the visual positioning accuracy is improved by multi-constraint dynamic feature points elimination; then, a tightly-coupled filter based on adaptive MCC for GNSS/INS/Vision is constructed to improve the reliability of the integrated system in urban traffic environment. The results of vehicle-mounted road experiments show that the vision-assisted based on multi-constraint dynamic feature points elimination can effectively improve the positioning continuity of the GNSS/INS integrated system, the positioning error is controlled within 2 m, the 3D RMSE is 59.73 cm, and the positioning accuracy is 15.44 % higher than the existing method. The positioning error of the adaptive filter based on MCC is controlled within 1.5 m in urban traffic environment, and the 3D RMSE is 48.56 cm, which improves the positioning accuracy by 13.16 % compared with the existing method.