Objective To overcome the limitations of experience-based parameter setting in orchard UAV spraying, which cannot adequately account for differences in fruit-tree size, canopy density, and operational objectives, a spray parameter recommendation method based on 3D point-cloud-derived structural features was developed.

Method Multi-view point clouds of 90 fruit trees were acquired with a Kinect V2 camera. Five structural features—tree height, volume, surface area, projected area, and light interception efficiency—were extracted, and K-means clustering was used to divide the trees into Class A (moderate volume and greater openness) and Class B (larger canopies with denser foliage). Nine spraying treatments were formed by combining three relative flight heights (2, 3, and 4 m) with three droplet-size classes (very coarse, medium, and very fine). Using droplet deposition rate, coverage density, deposition uniformity, and canopy penetration as evaluation indices, a recommendation rule was established based on linear weighted ranking and validated with independent samples.

Result For Class A, the combination of a 3 m relative flight height and very coarse droplets gave the best deposition rate and deposition uniformity, with an overall deposition rate of 41.3% and a coefficient of variation of 17.31%; better penetration into the middle and lower canopy was obtained at a relative flight height of 2 m with medium droplets. For Class B, the highest deposition rate, 40.2%, was achieved at a relative flight height of 3 m with very coarse droplets; the best uniformity was obtained at a relative flight height of 4 m with very fine droplets, with a coefficient of variation of 18.60%; and better penetration into the middle and lower canopy was achieved at a relative flight height of 3 m with very fine droplets. Validation results showed that the proposed method could output matched parameter combinations according to fruit-tree class and the priority assigned to spray-performance indices.

Conclusion The proposed method enables precise matching between fruit-tree structural class and spray parameters, and provides a reference for spray parameter recommendation and precision application in orchards.