Multiple Model Method for Aircraft Conflict Detection and Resolution in Intent and Weather Uncertainty

This paper presents a novel multiple model (MM) method for aircraft conflict detection and resolution (CDR) in intent and weather uncertainty for the next generation air transportation system. It is based on probabilistic MM aircraft trajectory prediction. Conflict detection is performed through the use of predicted probability of conflict (PC). An improved algorithm for estimating PC is proposed and compared with a commonly used method. Conflict resolution (CR) is formulated as a stochastic model predictive control (MPC) problem subject to a constraint on the predicted PC to guarantee safety of the optimized CR trajectories. The cost function to be minimized is tailored to the CR problem and includes the cost of performing evasive maneuvers as well as the average cost incurred by extra travel distance to the destination due to the maneuvers. An efficient algorithm for finding the optimal path (maneuver sequence), with respect to the maneuver cost, is proposed and extended to an overall CDR optimization algorithm for the total cost, in order to produce optimized conflict-free CR trajectories. The proposed search algorithm is essentially a generalization of an optimal list Viterbi algorithm (for unconstrained minimization through a trellis) to the constrained case of our MPC-based CDR problem. The capability and computational efficiency of the proposed MM-CDR method (including PC prediction, optimal search, and overall CDR) are evaluated via comprehensive simulation of a large variety of unmanned aerial vehicle 'sense-and-avoid' encounter scenarios, and are compared with existing methods in the literature.