High-precision RTK Positioning with Tilt Compensation: Data Fusion Algorithm

Abstract

ION GNSS+ 2021 Student Paper Award winner. Global navigation satellite system (GNSS) real-time kinematic (RTK) positioning with tilt compensation breaks the conventional positioning pattern with level-free and tilted surveys. The inertial measurement unit (IMU)-based solution uses mainstream tilted RTK principles, but it is disturbed by inertial navigation system (INS) alignment and degraded performance with increasing tilt angle. This solution has been applied to the GNSS receivers of some companies. However, there are few relevant literature reports, and technical details are not available. In this paper, we propose a low-cost and efficient IMU-based tilted RTK solution with relevant algorithms, including a rapid and accurate heading alignment method and zero velocity update (ZUPT) with lever arm compensation. The proposed heading alignment method calculates the initial heading from the angle between INS-indicated and RTK-indicated trajectories. An INS updating algorithm distinguished from conventional one is used to obtain INS-indicated trajectories by treating the pole movement as rigid body motion and discarding accelerometer measurement. ZUPT with lever arm compensation doesn’t require the pole to keep completely still. This ZUPT works when the pole tip touchs the ground. By validating the zero velocity of the pole tip, we can constrain the position error divergence and reduce the defects of the tilt angle. Then, accuracy analysis is performed, and explicit error equations are obtained. Experiments are carried out to validate the proposed solution, and the statistical results indicate that 1) the initial heading can be accurately determined to be 1.15? at a 98.2% confidence level within 2?3 seconds; 2) taking the ratio of horizontal error within 2 cm as a criterion, ZUPT with lever arm compensation can raise this ratio effectively even at a large tilt angle; 3) the horizontal position error of the proposed tilted RTK solution is within 2 cm at a 96.97% confidence level at a normal tilt angle.