An Evolutionary Computational Analysis of Tactical Controller Tool

Eurocontrol has introduced a Tactical Controller Tool (TCT), for use by Tactical Air Traffic Controllers (ATCs) who require trajectory conflict information over the next 5 to 8 minutes to help them detect conflicts in their sector. Simulation trials indicate that the safety benefits of TCT may be limited because of large numbers of False Alarms. There is a need to tune the system to identify the best look-ahead time to reduce nuisance alerts while retaining genuine conflicts generated by the system. In this paper we quantitatively investigate the performance of TCT for different look-ahead timings (using evolutionary computation to evolve complex air traffic conflict scenarios), and we investigate the patterns in conflict alerts raised by TCT that resulted in False Alarms. We find that a 6-minute look-ahead time leads to TCT generating fewest False Alarms. Flights in climb phase and with wide convergence angle contribute to a large number of False Alarms. TCT predicted conflicts that have duration of less than 45 seconds, and are on the boundary of 5 nm separation also lead to high numbers of False Alarms.