Adaptive Filter Design for UAV Navigation with GPS/INS/Optic Flow Integration

An integrated system of GPS and low cost Inertial Navigation System (INS) is often used to provide position, velocity and attitude (PVA) information for navigating Unmanned Aerial Vehicles (UAV). One drawback is that such systems can not provide the ground height information during landing and terrain following tasks which is essential for safety in near ground operations. Optic flow rate measurements can be used as additional observations in the data fusion process to augment the PVA estimation. Whilst this integration scheme is effective, further research has revealed that stochastic modelling uncertainty has a significant impact on its overall performance, especially when the stochastic characteristics of optic flow measurements are hard to define. To improve the filtering performance and to cope with the stochastic modelling uncertainties, in this research a covariance matching based adaptation algorithm has been implemented with the extended Kalman filter in a loosely coupled scenario. The proposed new integration scheme is evaluated with the field data collected from a UAV platform. Test results have showed clear performance improvements.