Planning for active mapping of crop fields using UAVs

National audienceAs more and more farmers are turning to drones to monitor the health of their crops and to assist them in decision making, there is an increasing interest in efficient mapping algorithms. Nowadays, the coverage path planning problem for crop fields using unmanned aerial vehicles (UAVs) is often solved offline with a pre-planned trajectory and the acquisitions are then compiled at the end of the flight. Such a non-adaptive approach may yield, for some reasons, sub-optimal maps which do not cover the whole target area or with quality defects.In this research, we propose a novel method based on metaheuristics to produce an optimised flight time trajectory to survey non-convex polygon areas in wind with a fixed-wing UAV. While offering better results than classical greedy approaches, this algorithm also proves to greatly reduce execution time. Thus, it can be used for online replanning tasks where the plan goals are deduced from a vision-based simultaneous localisation and mapping (SLAM), which exhibits areas where additional data must be gathered. This work is developed and validated within a realistic simulation framework