QC-MAINS: A Quality Control-based Multi-Antenna GNSS/INS Tightly Integrated Model

The integration of Global Navigation Satellite System (GNSS) and Inertial Navigation System (INS) has been widely used in civilian vehicular navigation, which can provide precise position and attitude for vehicle control and trajectory planning. Nevertheless, the accuracy of the GNSS/INS combined system inevitably diminishes when GNSS signals are frequently obstructed or interrupted, leading to an insufficient number of GNSS observations to mitigate INS drift. To improve the performance of the GNSS/INS system in such GNSS-compromised environments, nowadays the mainstream solution is further combining other sensors, i.e., Lidar, Camera or odometer. The crux of these approaches lies in introducing external pose information to supplement the missing GNSS signals. However, more sensors mean more complex system, as well as higher cost.

In many so-called GNSS-compromised environments (i.e., routes under trees, urban canyons), GNSS signals do not completely vanish but are characterized by reduced quantity and weakened quality. This raises the question: Has the GNSS information under harsh environments been completely utilized? The answer obviously is no. To realize the fully utilization of the GNSS information, this work studied a Quality Control-based Multi-Antenna GNSS/INS Tightly Integrated Model (QC-MAINS). The tight integration of INS and the multiply antennas located at different positions on a vehicle can explore potential GNSS observations as much as possible. Furthermore, the low-quantity GNSS observations can be detected and isolated through cross-checks with multiple antennas. Experiment results show that the horizontal position root-mean-square errors (RMSEs) are reduced by about 42, 54, and 48% for three harsh routes, respectively.