Research On Localization Algorithm Of Airport Enclosure Patrol Car Based On GIS And Environmental Perception's Fusion

With the development of civil aviation industry, the level of airport enclosure security has become higher and higher, developing mobile monitor station based on autonomous patrol vehicle is useful. Accurately perceiving airport enclosure environment and locating is the key of patrol car's self-navigation, but environmental suitability of single-sensor's traditional location method is not high. So it's meaningful to research vehicle's location of multi-sensor fusion based on airport enclosure environment.Firstly, single sensor's positioning error is analyzed and corrected. The localization principle and error sources of each sensor are analyzed. Meanwhile, single sensor's reliability module is built, paving the way for multi-sensor fusion.Secondly, a method of vision localization is raised based on Harris-SIFT feature extractors by improving SIFT to solve sensors blind zone problem on location. This method is a part of realizing self-organization location by establishing the region of interest, detecting Harris-SIFT characteristic points, doing pattern matching, calculating car's pose parameters with triangulation location and visual positioning method's confidence coefficient.Thirdly, a self-organization positioning algorithm based on perception dominance is proposed, combined with single sensor's confidence coefficient, building perception system,optimization layer and calculating perception dominance to choose the better way of multi-sensor information fusion at the moment. Finally, the result is shown. The larger, the better. This method can reduce position error when driving and improve vehicle's accuracy.Finally, a location algorithm combining GIS information and environment perception information is adopted. The results of self-organization positioning algorithm and electronic map data is contrasted to correct location information. Its longitudinal errors are reduced,making vehicle's pose information more precisely.Aiming at the algorithms mentioned in this paper, experiments on this vehicle platform has been done, the results show that it can improve patrol car's ability to adapt the environment.