Research on visual navigation technology of unmanned aerial vehicle landing

Abstract

Guide the UAV landing by using ground visual navigation system. Firstly, introduce the basic space intersection model: two videos capture sequence images of UAV and two theodolites record the azimuth and elevation of the optical axises of two cameras simultaneously. Then the position of UAV can be obtained by means of intersecting in space. Secondly, propose the adjustment model by introducing adjustment theory to improve the calculation precision. Finally, using kalman filter to improve the position accuracy further. The simulation model of visual navigation system is based on 3ds Max software. The simulation experiments show that, the basic space intersection model has high accuracy; the adjustment model can improve the accuracy in elevation direction; the position accuracy is further improved by introducing of kalman filter. Combining the above models and algorithms, the vision navigation system presented in this article meets the requirements of UAV landing.