Deterministic and Probabilistic Hybrid Control of Air Traffic Management

The increasing demand for air travel is stressing the current, mostly human operated ATM (air traffic management) system. It has been suggested that the enhanced automation in future ATM may alleviate some of this stress by improving the efficiency of the system and simplifying the task of the human operators. This improvement, of course, has to be achieved while maintaining (or ideally improving) the level of safety over the current system.

In ATM, safety is typically quantified in terms of numbers of conflicts, that is, situations where aircraft come closer to one another than a certain desired minimum distance. To prevent conflicts, ATM resorts to a two-stage process. In the first stage, conflict detection is performed; the positions of the aircraft in the future are predicted and compared to determine the possibility of conflict. Once a potential conflict has been detected, stage two "the conflict resolution stage" is invoked, to modify the plans of the aircraft.
Currently, all these functions are performed manually by the pilots and air traffic controllers (ATCs). Some partial automation tools are already available to assist the operators (for example, CTAS and TCAS). Conflict prediction and resolution are considered at three different levels of air traffic management process. The main characteristics of our contribution are the following:

(1) Probabilistic models are proposed for the aircraft position projection and for the validation of the proposed algorithms by Monte-Carlo simulation. The stochastic model for projecting the position of an aircraft in the future is simple and allows in principle fast computations, which makes it ideal for online conflict prediction. The validation model is more accurate than the prediction model, and therefore more difficult to compute with; this is not a major concern, however, as it is only used offline.

(2) A detection algorithm based on the proposed prediction model is introduced. The prediction model produces probability distributions for the future positions of the aircraft, which are used to construct a probabilistic measure of the criticality of the situation. If the measure exceeds a certain threshold, a conflict is declared.

(3) The computational issues involved in the application of the proposed conflict altering system are addressed by resorting to randomized algorithms. The advantage of randomized techniques is that they tend to be computationally more efficient. Moreover, the computational load does not significantly increase in the 3D case with respect to the 2D case. They also provide analytical bounds on the accuracy of the approximation involved.

*This project is not officially supported through CITRIS funds, but the faculty and topical affiliations are sufficiently strong that it is listed here for referral and convenience.