A New Triple Filtering Algorithm and Its Application for Aerial GNSS/INS-Integrated Direct Georeferencing System
Autor: | Yu Wu, Keyi Xu, Qusen Chen, Leilei Li, Xiangdong An |
---|---|
Jazyk: | angličtina |
Rok vydání: | 2021 |
Předmět: |
Article Subject
Computer science Stochastic modelling Filter (signal processing) Computer Science::Robotics Extended Kalman filter Control and Systems Engineering Inertial measurement unit GNSS applications T1-995 Observability Electrical and Electronic Engineering Allan variance Instrumentation Algorithm Technology (General) Inertial navigation system |
Zdroj: | Journal of Sensors, Vol 2021 (2021) |
ISSN: | 1687-725X |
DOI: | 10.1155/2021/6527356 |
Popis: | A global navigation satellite system and inertial navigation system- (GNSS/INS-) integrated system is employed to provide direct georeferencing (DG) in aerial photogrammetry. However, GNSS/INS suffers from stochastic error, strong nonlinearity, and weak observability problems in high dynamic or less maneuver scenarios. In this paper, we proposed a new triple filtering algorithm for aerial GNSS/INS integration. The new algorithm implements filtering in the sequence of forward, backward, and forward directions. Each filter is initialized by a previous filter to get a quick convergence, and the final result is combination of the last two filtering to smooth error. The proposed triple filtering strategy avoids inaccuracy in the 1st forward filtering when the system has not reached convergence. Moreover, it facilitates engineering implementation because backward filtering can employ the same equations with forward filtering. To assess stochastic error of the inertial measurement unit, the Allan variance method is used and abbreviated stochastic model is built. A real aerial testing is conducted, and the result indicates that DG can achieve horizontal accuracy of 5 cm by the proposed algorithm, which has 63% improvement compared to standard extended Kalman filter. |
Databáze: | OpenAIRE |
Externí odkaz: |