Inferring Inertial Navigation Errors from SAR Image Distortions Using a Convolutional Neural Network

<title>Abstract</title> Unmanned aerial vehicles often rely on the Global Positioning System (GPS) for navigation. GPS signals, however, are very low in power and can be easily jammed or otherwise disrupted. This paper presents a method for estimating the navigation errors present at the beginning of a GPS-denied period using data from a synthetic aperture radar (SAR) system. These errors are estimated by comparing an online-generated SAR image with a reference image obtained a priori. The distortions relative to the reference image are exploited by a convolutional neural network to learn the initial navigation errors, which can be used to recover the true flight trajectory throughout the synthetic aperture. The proposed neural network approach is able to learn to predict the initial errors on both simulated and real SAR image data.