Zsedrovits et al. This strategy can be scaled by the length of course angle rate decrease over time. Exact proof of avoidance guarantee cannot be provided, but Monte Carlo simulations can be done with different intruder sizes and velocities. Preliminary flight test results are presented in the next section. Detailed test and tuning are part of future work. Possible performance metrics can be the missed detection (MD) and false alarm (FA) rates. MD means characterization of a MAC/NMAC scenario as a non-collision one, while FA means the avoidance of a non-threatening intruder. PRELIMINARY FLIGHT TEST RESULTS In these flight test results, the E-flite Ultrastick 25e (1.2 m wingspan 2 kg) and Sindy (3.5 m wingspan 12 kg) aircrafts operated by SZTAKI Systems and Control Laboratory are used as intruder and own aircrafts, respectively, applying an onboard autopilot developed in-house [26]. The described camera system (with USB 3.0 1280 × 960 cameras) are mounted on Sindy and the described image processing, intruder detection, and tracking and decision al- Table 2. Required Course Angle Rates to Hold the Intruder at Tbeta Target Angles (Intruder Path Originally Parallel with Own Path (Beta = 0°)) Beta [deg] Vi [m/s] Vo [m/s] om0 [deg/s] Tbeta [deg] distance [m] max(om) [deg/s] 0 14 15 9 50 0 72 0 14 15 9 50 -25 37 0 14 15 9 50 -50 24 0 14 15 9 50 -75 17 0 14 15 9 50 -100 14 0 14 15 9 90 0 62 0 14 15 9 90 -25 34 0 14 15 9 90 -50 23 0 14 15 9 90 -75 18 0 14 15 9 90 -100 14 Table 3. Required Course Angle Rates to Hold the Intruder at Tbeta Target Angles (Intruder Path Crosses Own Path (Beta = 45°)) Beta [deg] Vi [m/s] Vo [m/s] om0 [deg/s] Tbeta [deg] distance [m] max(om) [deg/s] 45 14 15 9 50 0 68 45 14 15 9 50 -25 38 45 14 15 9 50 -50 25 45 14 15 9 50 -75 19 45 14 15 9 50 -100 15 45 14 15 9 90 0 65 45 14 15 9 90 -25 37 45 14 15 9 90 -50 26 45 14 15 9 90 -75 20 45 14 15 9 90 -100 16 SEPTEMBER 2016 IEEE A&E SYSTEMS MAGAZINE 25