TRANSPORT Mathematical Model of Errors of Odometry… 31 1. INTRODUCTION UAV navigation in GPS-denied environment can possibly rely on low-cost dead reckoning system like inertial navigation system (INS) and therefore needs continuous correction from external sources. Promising variant of such correction is use of visual information available onboard. Realization of correlation extreme principle is based on comparison of current image (or frame) with template geo-referencing image and on finding coordinates of UAS location. Visual data has rich informativity and provides the extraction of such flight and navigation parameters like coordinates, velocity, orientation angles, etc. Their values can be used further in data fusion schemes to minimize the error accumulation in INS. Errors of visual correlation extreme systems may influence significantly the accuracy and efficiency of data fusion and therefore require strong attention and research. Sources of errors during navigation process by visual observation: – noises caused by technical imperfection of camera (fixed pattern noise, dark current noise, shot noise, amplifier noise and quantization noise [1]) An estimate of noise level can be obtained by noise level function [2] for a given image, which predicts the overall noise variance at a given pixel as a function of its brightness (a separate function is estimated for each color channel). – cartographic errors due to inaccuracy of template images, operative changes of environment, use of different reference systems, etc [3]. – measurement method errors because of coordinate system conversion, inaccuracy of matching process [4], etc. – Among others it is also necessary to take into account the type of used visual navigation: absolute (or geo-referencing) [5] and relative (so called visual odometry [6]). 2. PREVIOUS RESEARCHES ANALYSIS Random component of visual CENS errors in most researches [7] is represented as additive Gaussian white noise with zero mean or possible as a mix of Gaussians. Constant component is used to describe measurement-method errors to be calibrated during pre-flight procedure. A geo-referencing system for absolute UAV positioning was developed in [7]. The position reference is expressed as a standard measurement equation, making it easy to incorporate into any sensor fusion framework. The system makes use of environmental classification and rotation invariant template matching, making it robust to variations in the operational environment as well as errors in Mathematical Model of Errors of Odometry and Georeferencing Channels in Visual Correlation Extreme Navigation Maryna Petrivna Mukhina, Volodymyr Petrovych Kharchenko National Aviation University, Ukraine The mathematic model of errors in correlation with the extreme navigation system (CENS) is developed basing on odometry and geo-referencing channels. The realization of the model is done in Simulink, and based on regular and random components of additive noise. The results of simulations prove accumulation of errors for odometry errors and its mitigation in case of geo-referencing in periods of correction. Keywords: correlation extreme navigation, visual odometry, geo-referencing, feature matching, cartographic errors, dead reckoning errors.