Flight Demonstration of Provably Safe Closely Spaced Parallel Approaches

In poor visibility, paired approaches to airports with closely spaced parallel runways are not permitted, thus halving the arrival rate. With Global Positioning System technology, datalinks and cockpit displays, this could be averted. One important problem is ensuring safety during a blundered approach by one aircraft. This is on-going research. A danger zone around the blunderer is required. If the correct danger zone could be calculated, then it would be possible to get 100% of clear-day capacity in poor-visibility days even on 750 foot runways. The danger zones vary significantly (during an approach) and calculating them in real time would be very significant. Approximations (e.g. outer bounds) are not good enough. This paper presents a way to calculate these danger zones in real time for a very broad class of blunder trajectories. The approach in this paper differs from others in that it guarantees safety for any possible blunder trajectory as long as the speeds and turn rates of the blunder are within certain bounds. In addition, the approach considers all emergency evasive maneuvers whose speeds and turn rates are within certain bounds about a nominal emergency evasive maneuver. For all combinations of these blunder and evasive maneuver trajectories, it guarantees that the evasive maneuver is safe. For more than 1 million simulation runs, the algorithm shows a 100% rate of Successful Alerts and a 0% rate of Collisions Given an Alert. The evaluation of the proposed scheme was demonstrated with two different types of test platforms, the Stanford DragonFly unmanned aerial vehicles and the Boeing testbed of F-15 and T-33 aircraft. The experiment involved two aircraft, where one aircraft is assigned as pursuer and the other as evader. The algorithm proved the guaranteed safety (maintaining a minimum separation distance) for every single flight test performed, and demonstrated that it can run it in real-time.