Pollination is a critical determinant of date palm productivity across tropical and subtropical regions. However, manual pollination remains labor-intensive, costly, and challenging to scale during staggered flowering periods that require multiple interventions. This study introduces a prototype and simulation-based framework for AI-enabled, drone-assisted pollination of date palms, developed to improve precision, scalability, and sustainability in agricultural operations. The system integrates YOLOv5 and YOLOv8 models for automated inflorescence detection and supports both semi-autonomous and fully autonomous flight modes that link perception to pollen-spraying actuation. A carefully prepared and manually labeled dataset was used to train and evaluate these models across varied illumination and canopy conditions. Model performance, expressed in mean Average Precision (mAP), along with operational metrics such as labor, time, and pollen efficiency, demonstrated substantial improvements over traditional methods. The proposed design achieved an estimated 80% reduction in labor and 97% reduction in pollen usage compared to manual pollination, highlighting its potential as a scalable and cost-effective alternative. While agronomic parameters such as fruit set percentage (FSP), yield, and quality were not directly measured, this work establishes a decision-support and prototype baseline for future large-scale field trials under diverse environmental conditions, including. wind, humidity, and canopy density.
AlRaeesi et al. (Sun,) studied this question.