GNSS PPP-RTK tightly coupled with low-cost visual-inertial odometry aiming at urban canyons

• Published in: Journal of geodesy Vol. 97; no. 7
• Main Authors: Wang, Feng, Geng, Jianghui
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.07.2023; Springer Nature B.V
• Subjects: Canyons; Components; Earth and Environmental Science; Earth Sciences; Geodetics; Geophysics/Geodesy; Inertial navigation; Navigation; Navigation satellites; Navigation systems; Navigational satellites; Original Article; Statistical methods; Unmanned aerial vehicles; Urban areas; Urban environments; Vision; Urban canyons navigation; PPP-RTK/INS/Vision tightly coupled integration; PPP-RTK
• ISSN: 0949-7714; 1432-1394
• DOI: 10.1007/s00190-023-01749-7

High-precision positioning is essential for emerging mass-market applications such as autonomous driving, unmanned aerial vehicles and intelligent transportation in complex urban canyons. Without global navigation satellite system (GNSS) reference stations, the real-time kinematic precise point positioning (PPP-RTK) technique is widely used in these fields because of its fast and accurate absolute positioning capability. However, in complex urban canyons, the performance of PPP-RTK deteriorates dramatically due to the satellite signal obstruction and interference. In contrast, inertial navigation system (INS) and visual navigation can provide precise relative pose estimation, but they suffer from serious error accumulation. Therefore, there is a strong complementary advantage between GNSS and INS/Vision navigation and positioning. In this study, a tightly coupled PPP-RTK/INS/Vision integration model is developed, aiming at improving the performance of PPP-RTK in terms of precision and availability to achieve robust positioning in complex urban canyons. Vehicle-borne experiments with different typical urban scenarios were carried out. We found that the positioning precision, availability and the ambiguity fixing rate of PPP-RTK can be significantly improved by tightly coupling with INS and Vision. The positioning precision of tightly coupled micro-electro-mechanical system (MEMS)-based PPP-RTK/INS/Vision is 6 cm, 5 cm and 10 cm in the east, north and up components, respectively, with an ambiguity fixing rate of 83.6% in typical urban environments, which are comparable to the statistics of a tactical-based PPP-RTK/INS tight integration. Even in complex urban canyons, the positioning precision of tightly coupled MEMS-based PPP-RTK/INS/Vision solution can still reach 11 cm, 7 cm and 13 cm in the east, north and up components, respectively, exhibiting improvements of over 90% compared to GNSS-only PPP-RTK. These results indicate that PPP-RTK tightly coupled with low-cost MEMS-IMU and monocular vision can provide high-precision and high-reliability navigation and positioning solutions even in complex urban environments.