Decentralized Cooperative Trajectory Planning of Multiple Aircraft with Hard Safety Guarantees

This paper presents a framework for provably safe decentralized trajectory planning of multiple (autonomous) aircraft. Each aircraft plans its trajectory individually using a receding horizon strategy based on mixed integer linear programming (MILP). A constrained, inertial, first-order linear model is used to capture the dynamics and kinematics of the vehicle. Safety is guaranteed by maintaining, at each time step, a dynamically feasible trajectory for each aircraft that terminates in a loiter pattern. Conflicts between multiple aircraft are resolved in a sequential, decentralized fashion, in which each aircraft takes into account the latest trajectory and loiter pattern of the other aircraft. Besides maintaining feasibility, if the problem is too complex to be solved within the time constraints of a realtime system, this approach also provides an a priori safe rescue solution consisting of the previous trajectories and individual loiter patterns. Several examples of conflict situations resolved by the proposed method are presented.