Detect and Avoid Tracker Augmentation (DAATA)

Detect and avoid (DAA) systems are combinations of sensors, trackers, and avoidance algorithms. The sensors provide surveillance information to the DAA trackers which in turn provide the location and velocity of nearby aircraft to the DAA’s alerting and guidance algorithm.  Advanced DAA tracker capabilities, such as those developed by AFRL, can further estimate the traffic’s horizontal turn rate, but cannot reliably predict upcoming turn-ins, roll-outs or changes in vertical profiles.  This limitation is exacerbated when applying DAA in the terminal area where traffic tends to flow in geographically prescribed patterns. Applying the current Transit DAA tracker for departure or arrival operations is expected to lead to elevated nuisance/late alert rates, flow interruptions and decreased safety margin in the terminal area. Arrival and departure traffic tend to follow patterns. These patterns are embedded in historical data which can be analyzed and extracted to improve DAA tracker prediction of future aircraft location and intent in terminal areas. As such, Bihrle Applied Research (BAR), proposes the use of machine learning to extract these embedded patterns to enhance and provide aircraft intent to tracker outputs. The DAA Tracker Augmentation (DAATA) system will take current tracker outputs, assess these outputs with respect to historical data, statistical modeling of the terminal airspace, and known terminal airspace procedures, and provide enhance DAA alerting and guidance to CA systems and AVOs. By design, DAATA will be sensor, tracker, and CA system agnostic, therefore applicable in a variety of deployment scenarios. BAR anticipates that there is enough underlying structure in the terminal airspace trajectory data to both categorize the aircraft trajectories as arrivals, departures, or transits, and predict where the aircraft will be in the future with an accompanying probability of accuracy. The structure in the data content that is expected is related to the procedures for arrivals and departures. Because of this, it is anticipated that a machine learning model will need to be trained for every deployment scenario and to be regularly updated as arrival and departure procedures change.  However, it is anticipated that the machine learned categorization will be agnostic to deployment scenarios as it is examining the overall shape of the aircraft trajectory and not predicting positions in time.  While this effort is focused on the development of DAATA for terminal areas, DAATA technologies can be applied to other situations and use cases. For instance, application to other congested airspaces like Carrier Controlled Airspace is a natural extension of DAATA. Additionally, with the push from the Air Force in its Agility Prime program and industry, application of DAATA technologies to Urban Air Mobility (UAM) settings could improve the overall safety and integration of AUAS into the urban airspaces.