Sammendrag
In this paper, a distributed linear MPC approach to solve the trajectory planning problem for rotary-wing UAVs is proposed, where the objective of the UAV system is to form a communication network to multiple targets with given radio communication capacities. The approach explicitly incorporates constraints on radio communication path losses, computed by using SPLAT! that is able to take into account terrain models and antenna locations. In order to enhance the online optimization, at each time sample the terrain below each UAV and the communication path losses are approximated with linear functions of the spatial coordinates. This leads to linear MPC sub-problems, which are solved by using convex quadratic programming. An algorithm for automatic initialization and optimal reconfiguration of the communication topology in case of failures or severe radio path loss due to e.g. channel fading, is proposed. The communication network that is provided by the UAVs is considered to be a payload communication capacity that is normally independent of the command and control datalink used to control the UAVs. The performance of the distributed linear MPC trajectory planning and the reconfiguration algorithm is studied on two simulation cases with four UAVs and two targets.