Adaptive Kalman filtering for integration of GPS with GLONASS and INS

In an integrated kinematic system, the Kalman filter is commonly used to integrate the data from different sensors (such as GPS/GLONASS and INS) for precise positioning. Reliable Kalman filtering results rely heavily on the correct definition of both the mathematical and stochastic models used in the filtering process: Whilst the mathematical models for various positioning measurements are (sufficiently) known and well documented in the current literature, stochastic modelling is not trivial, in particular for real-time applications. In this paper, a newly developed adaptive Kalman filter algorithm is introduced to directly estimate the variance and covariance components for the measurements. Example applications of the proposed algorithm in GPS/GLONASS kinematic positioning and GPS/INS integration are discussed using test data sets. Test results show that the proposed algorithm can improve the performance of the filtering process.