A Model-Based Drift Correction Control for UAV in GNSS-Degraded Environments

Among its many applications, Global Navigation Satellite System (GNSS) systems made an entrance into the Unmanned Aerial Vehicle (UAV) field with positioning and navigation. These applications usually require high positioning accuracy out of safety considerations. Moreover, UAV applications are often found in dense urban or forested areas with poor reception conditions for each available satellite signal, and, thus, degrades the overall accuracy of GNSS positioning. In particular, the position accuracy deteriorates in the face of environmental obstructions as services and multi-path receptions become unavailable. Our works consist of two parts. The first is a nominal Linear Quadratic Gaussian (LQG) controller that estimates the states and tracks the reference trajectory without GNSS drift. The LQG control determines an output feedback law that is optimal in minimizing the expected value of a quadratic cost criterion when the output measurements are corrupted by Gaussian white noise. However, when the reception state is Non-Line Of Sight (NLOS), the Gaussian distribution assumption is no longer valid. Therefore, the latter part of our work contributes a robust controller H
∞
that estimates the drift and corrects the trajectory accordingly. H
∞
control applies classical loop-shaping concepts to the multivariable frequency response for good robust performance. Given the estimate of GNSS drift as an input, H
∞
control not only improves the overall robustness of the control loop but also corrects the drift to converge ground truth to the reference trajectory.