Path Planning for a UAV Performing Reconnaissance of Static Ground Targets in Terrain

In this article we consider a path planning problem for a single fixed-wing aircraft performing an ISR (Intelligence Surveillance Reconnaissance) mission using EO (ElectroOptical) camera(s). We give a mathematical formulation of the general aircraft visual reconnaissance problem for static ground targets in terrain and show that, under simplifying assumptions, it can be reduced to a variant of the Traveling Salesman Problem which we call the PVDTSP (Polygon-Visiting Dubins Traveling Salesman Problem). We design a genetic algorithm to solve the PVDTSP and validate it in a Monte Carlo numerical study. For fixed computation time, the genetic algorithm produces reconnaissance tours on average nearly half the length (and thus can be flown in half the time) of those delivered by a random search. The modular design of the genetic algorithm allows it to easily be extended to handle realistic assumptions such as wind and airspace constraints.