Image-Aided Position Estimation Based on Line Correspondences During Automatic Landing Approach

This paper investigates the current techniques for vision-aided navigation procedures which can be applied on aircraft during their landing approaches. A feature based algorithm which relies on line and point correspondences is introduced. It provides an unambiguous 3D position vector with respect to the desired runway in terms of the required units, e.g. in meters. Furthermore, accuracies of the proposed procedure are discussed and a simple error model is presented which provides standard deviations of the estimated position vector. The presented results are gathered by simulated and real flight experiments performed on the research aircraft Dornier 128-6.