The expectation to the drone which is a small unmanned plane spreads, and the drone attracts attention as most advanced technology. It is expected in particular automatically running of the drone. The GPS is important to catch the position information of the moving object exactly, but the difficult case that is the acquisition gets up by weather and the state of the electric wave and the position relation of the satellite. We propose method to learn a path to let goal position go around a drone in this paper from the starting point without GPS. We let you fly a drone several times and fly it from the departure place to the destination at first. We learn flight path based on a picture and the flight recording at the time of the flight of the drone from several times of driving. We get a basis function suitable for learning from the picture at the time of the navigation for self-organization and classify image sequence from a remarkable feature to look it to image sequence. We learn operation to fly a drone in goal position without GPS from image sequence and control. We actually show the effectiveness by letting a drone fly from the starting point to an objective in the study room.