Optimization of coverage mission for lightweight unmanned aerial vehicles applied in crop data acquisition

摘要

The crop data acquisition with unmanned aerial vehicles is a popularized alternative to manage the agricultural processes, due to data emerging from portable sensors for image-gathering. Nevertheless, most unmanned aerial vehicles for data acquisition excess the cost of hundreds of dollars, making them inappropriate for small agricultural producers. In this paper, we proposed to achieve crop data acquisition using a Lightweight Unmanned Aerial Vehicle (LUAV), available at a reasonable cost. However, a LUAV has less flight time and robustness than the professional vehicles. To overcome the limitations, we designed a LUAV agent with the goal of optimizing coverage paths using a heuristic strategy in known areas. The path to follow can be selected from three algorithms, Wavefront, Dijkstra or Spiral, which are compared to define an option for the crop under study. A second goal is to improve the LUAV robustness, which was resolved from planning by selecting the start of the coverage mission in order to the flight lines cross the direction of the wind. We complemented the robustness of outdoors positioning using a Kalman Filter extension to specify movements during missions. Finally, using an AR Drone 2.0 quadcopter, we developed a prototype of the LUAV agent to obtain the mosaic of a grass crop. The results respect to optimized coverage mission showed that the Spiral algorithm with a Backtracking technique and avoiding areas of little interest, got the balanced score between revisits, turns, coverage percent and traveled distance. About the LUAV robustness in the presence of wind, the results stated an error of less than 2 m, considered acceptable for image-acquisition purposes. The developed work is simple but effective, and makes evident the viability for that any LUAV type can support the precision agriculture processes in favorable costs. (C) 2020 Elsevier Ltd. All rights reserved.