Previous Abstract Return to Session D3 Next Abstract

Session D3: GNSS Augmentation and Robustness for Autonomous Navigation

Short Baseline Real-time Kinematic Method using Moving Reference Antenna Array
Sae-Kyeol Kim and Euiho Kim, Hongik University
Alternate Number 2

Precise centimeter-level relative positioning between UAVs is essential for operation of multiple UAVs without collision and has often been obtained from using Global Navigation Satellite System (GNSS) Real-Time Kinematic (RTK). If the integer ambiguities of the carrier phase measurements in RTK method are accurately resolved, a precise navigation solution can be derived with centimeter-level accuracy. However, in the case of a moving baseline RTK problem, more GNSS measurements and time are required for correct to ambiguity resolution compared to a fixed baseline RTK. To have more number of measurements in an RTK process, RTK with antenna array fixed on ground has proved to be effective. The antenna array RTK technique uses accurate baseline information between ground-fixed reference antennas to resolve integer ambiguities, and improvement of positioning accuracy and ambiguity resolution success rate has been demonstrated through numerical results. In this paper, we applied the antenna array RTK concept to a moving baseline RTK problem such that flying UAVs are used as a reference antenna array instead of a ground-fixed one. For that purpose, we changed the previous least-squares based antenna array RTK formulations to a Kalman filter based formulations, which performs better with the measurements from the moving UAVs. To test the performance of the least-squares based and Kalman filter based antenna array RTK, we performed flight tests using multiple UAVs. The test results showed that our approach performs better than a conventional moving baseline RTK and the least square based antenna array RTK. The 3D-root mean square (RMS) position error of our Kalman filter based antenna array was 0.10m, while the RMS error of the conventional least square based moving baseline was 1.23m, the RMS error of the conventional Kalman filter based moving baseline was 0.35m, and the RMS error of the least square based antenna array was 0.83m. We concluded that antenna array particularly effective when there are 6 or less satellites in view.



Previous Abstract Return to Session D3 Next Abstract