UAV path planning for maximum visibility of ground targets in an urban area

Multiple moving targets in an urban area are to be tracked simultaneously by unmanned aerial vehicles. It is assumed that the moving targets try to avoid the camera field of view of the aircraft by changing their velocities and/or hiding behind buildings. The number of aircraft is much smaller than the number of targets, in general. In order to track as many targets as possible, firstly the targets are grouped into a number of subgroups by maximising the modularity, which is solved efficiently by the power iteration. Secondly, circular optimal paths are assigned to maximise the visibility of the area, given shapes and locations of the ground obstacles, where the computational complexity is reduced using a novel random sampling method. Finally, the aircraft transition paths from the current positions to the desired path are obtained by solving a discrete minimum weighted path length problem.