An Optical-Flow-Assisted Navigation Method for UAVs Using Sage-Husa EKF

Small unmanned aerial vehicles (UAVs) are usually equipped with low-cost global positioning systems (GPS)/inertial integrated navigation systems. In the GPS-denied environment, the low-cost inertial navigation system cannot meet the navigation requirements due to information divergence. This paper proposes an optical-flow-assisted navigation method for UAVs. Firstly, an improved sum of absolute difference (SAD) block matching algorithm is proposed to calculate the optical flow fields. Then, the calculated results are combined with range information to calculate the optical flow line-of-sight (LOS) angular velocity. Finally, the Sage-Husa EKF is used to fuse the inertial and optical flow information to suppress the divergence of inertial navigation information. Different matching criteria are used to calculate the optical flow field for the images collected by the vision sensors. The calculation results show that the improved SAD block matching algorithm can balance the accuracy and real-time performance of the calculation. The Sage-Husa EKF is used for the offline fusion of UAV sensor data. By comparing the results with those obtained by classical EKF method, this method proves to be able to suppress the filter divergence occurred in the information fusion process. The mean squared error (MSE) of the filtering results is improved.