Abstract. The paper proposes an algorithm for planning routes for overflighting agricultural fields with obstacles to solve precision farming problems. The algorithm can be applied in both processing and field monitoring tasks. Unlike classical methods, which are limited to simple zigzag traversal (Zamboni) and elementary perimeter traversal of obstacles, this algorithm takes into account the presence of a fleet of heterogeneous drones (different type, range, cost, speed) and a moving ground platform that provides energy and necessary resources for the process of overflight. The drones take off and land at various points along a road that typically wraps around a field. The key innovation is a two-stage optimization procedure: first, a random set of field partitions is generated into several sub-multiples with specified area fractions (taking into account internal obstacles), and then, for the optimal partition, a genetic algorithm is run to optimize the overflight parameters (flight angle, entry point, composition and order of drone launches, and platform route). Optimization is achieved through a more localized traversal of certain parts of the field (each area is served by a suitable type of drone), as well as flexible movement of the ground platform, reducing useless flights. Numerical experiments show that, depending on the size of the obstacle and the size of the field, a 12-15% reduction in the cost of overflights is achieved. The final part of the paper discusses the prospects of the solution development, including consideration of 3D terrain, dynamic factors (changing weather conditions, drone stopping for technical reasons) and automatic collision prevention on intersecting route sections.

Keywords: Coverage algorithm, drones, genetic algorithm, flight planning, artificial intelligence.