On GNSS/MIMU integrated navigation system based on observability theory

An improved Kalman filter method which meets the basic requirements of accuracy and real-time character is presented to resolve the real-time navigation problems in certain situations such as UAV or carrier-UAV precision landing. An on-the-move alignment method is adopted for the low-cost and low-accuracy MEMS-INS, which utilizes the information of GNSS available to obtain the initial value for the fine alignment stage using velocity plus position update matching method. The fine alignment and data fusion of the integrated system is realized by utilizing the improved algorithm. It processes the variables according to their observability, dividing them into two categories-the observable variables and the unobservable ones. And these two kinds of variables are processed differently, i.e. the former would be updated by Kalman filter while the latter would inherit the previous value. As a result, the vehicle experiment indicates that the integrated navigation system equipped with low-cost and low-accuracy MEMS-INS, not only maintains high-accuracy of the GNSS RTK, but also improves continuity and reliability of navigation(about 0.15m when GNSS available; under 3.5m when GNSS outages for 5 sec.).