Landmark-aided localization for air vehicles using learned object detectors

This research presents a real time method to localize an aircraft without GPS using fixed landmarks observed from an optical sensor. The objective is to achieve practical navigation performance using available autopilot hardware and a downward pointing camera. Computer vision cascade object detectors are trained to detect selected landmarks prior to a flight. The method also explores aircraft localization using roads between landmark updates. During a flight, the aircraft navigates with inertial measurements and obtains measurement updates when landmarks are detected. Inertial measurements and information extracted from the aircraft’s camera images are combined into an unscented Kalman filter to obtain an estimate of the aircraft’s position and wind velocities.