Decomposition-based mission planning for fixed-wing UAVs surveying in wind

This paper presents a new method for planning fixed-wing aerial survey paths that ensures efficient image coverage of a large complex agricultural field in the presence of wind. By decomposing any complex polygonal field into multiple convex polygons, the traditional back-and-forth boustrophedon paths can be used to ensure coverage of these decomposed regions. To decompose a complex field in an efficient and fast manner, a top-down recursive greedy approach is used to traverse the search space to minimize the flight time of the survey. This optimization can be computed fast enough for use in the field. As wind can severely affect flight time, it is included in the flight time calculation in a systematic way using a verified cost function that offers greatly reduced survey times in the wind. Other improved cost functions have been developed to take into account real-world problems, for example, No-Fly Zones, in addition to flight time. A number of real surveys are performed to show the flight time in wind model is accurate, to make further comparisons to previous techniques and to show that the proposed method works in real-world conditions providing total image coverage. A number of missions are generated and flown for real complex agricultural fields. In addition to this, the wind field around a survey area is measured from a multirotor carrying an ultrasonic wind speed sensor. This shows that the assumption of steady uniform wind holds true for the small areas and time scales of an unmanned aerial vehicle aerial survey.