Task Assignment and Motion Planning for Multiple UAVs Tracking Multiple Targets in Urban Environments

Abstract

Listen

Translate article

We propose a combined task assignment and motion planning algorithm for a team of heterogeneous unmanned aerial vehicles (UAVs), tracking multiple predictable ground targets in a known urban environment. The proposed algorithm is distributed, thus the computational workload is divided between the team members. This enables the execution of the algorithm in relatively large teams of UAVs and multiple targets. The solution methodology involves finding visibility regions, from which a UAV can maintain line of sight to each target during the scenario; and restricted regions, in which a UAV can not fly, due to the presence of buildings or other airspace limitations. These regions are then used to pose a combined task assignment and motion planning optimization problem, in which each UAV’s cost function is associated with its location relative to the visibility and restricted regions, and the tracking performance of the other UAVs in the team. A distributed co-evolution genetic algorithm is derived for solving the optimization problem. The proposed solution is scalable, robust, and computationally parsimonious. The viability of the algorithm is demonstrated in a Monte Carlo study, using a high fidelity simulation test-bed incorporating a visual database of an actual city.