A vision based parameter estimation for an aircraft in approach phase

Abstract

Landing and takeoff are critical phases of an aircraft flight. In recent years automation has taken over some those crucial functions, especially the landing. Aircraft such as Airbus A220 and many business jets are capable of landing autonomously, performing a rollout and complete stop without pilot intervention. These autoland systems require extensive ground-based navigational aids, which increase the cost of airport construction and maintenance. Approaches such as Localizer Performance with Vertical (LPV) guidance, which is not dependent on extensive ground-based navigational aids, provides guidance using GPS position but the system has a decision height of 200 ft. Presented here will be a first step of a broader investigation to use computer vision to develop necessary aircraft-based technology that enables fully autonomous landing by augmenting the LPV approach and providing robust estimates of aircraft states, thus reducing the decision height to 0 ft. Work presented here uses runway data from National Airspace System Resource (NASR) System combined with camera properties to estimate the states of a fixed wing aircraft relative to a runway during the approach phase of flight. The algorithm was able to estimate the position of aircraft within 50 ft during the final phases of approach.