Collision-free UAV formation flight using decentralized optimization and invariant sets

We consider the problem of formation flight for a set of unmanned air vehicles (UAV). We propose a decentralized control design procedure which guarantees collision avoidance and constraint fulfillment. The control design is based on a decentralized receding horizon control (RHC) scheme. Vehicle collision avoidance is ensured by considering a collision-free emergency maneuver which is implemented when feasibility of the decentralized RHC scheme is lost. Bounds on speed and accelerations are computed off-line using simple polyhedral invariant set computations. Such bounds guarantee that the implementation of the emergency maneuver leads to collision-free trajectories. The proposed decentralized control scheme is formulated as mixed-integer linear programs of small sizes which can be translated into equivalent piecewise affine state-feedback controllers. These controllers can be implemented in real-time once the corresponding look-up tables are downloaded to the hardware platform of the UAVs.