Monocular image parameter-based aircraft sense and avoid

This paper deals with the problem of monocular image parameter-based sense and avoid. It considers image parameters as decision variables and selects decision thresholds related to collision and non-collision scenarios. The main contri- bution is non-heuristic threshold selection. Another contribution is the characterization of possible intruder threats with size- speed curves given in closed form formulae. This makes it possible to avoid the use of lookup tables. The overall decision and avoidance concept is evaluated in a software-in-the-loop simulation campaign considering threats ranging from small UAV to large airliner. The miss detection rate of the method is zero which is an excellent result, however the false alarm rate is high. The causes of this are pointed out and targeted as further developments. Finally, promising results are presented executing the method for real camera images.