This paper implements and demonstrates an optimal positioning and trajectory planning algorithm and framework for multiple unmanned aerial vehicles (UAVs) to improve the communication network of static ground nodes with unknown locations. In particular, a two-stage framework is proposed. In the first phase, UAVs first localize the ground nodes using received signal strength measurements. Next, the UAVs transition to acting as relays between the ground nodes once their locations have been determined, within a certain level of tolerance. In the first stage, the UAVs assigned to the search task cooperatively plan their trajectories in order to maximize the determinant of the Fisher Information Matrix for all of the ground nodes. In the second stage, the UAVs assigned to the relay task cooperatively plan their trajectories in order to minimize the cost of the minimum spanning tree of the graph that models the overall network. In both stages, the UAVs plan in a distributed manner with a receding horizon approach. A discrete genetic algorithm is used to find the optimal control inputs while considering the collision avoidance and dynamic constraints of the fixed-wing UAVs. The UAVs switch from one task to another when their base station determines that the estimate for a ground node has reached an acceptable level of uncertainty and that the ground node is disconnected from the network. Demonstrations of this framework with a single and multiple targets and UAVs are presented along with analysis of the performance of the trajectory planning in improving estimates of target locations and in improving the connectivity of the ground network.
|Original language||English (US)|
|Title of host publication||AIAA/IEEE Digital Avionics Systems Conference - Proceedings|
|Publisher||Institute of Electrical and Electronics Engineers Inc.|
|State||Published - Oct 11 2020|