Autonomous Drone-Based Pollination System Using AI Classifier to Replace Bees for Greenhouse Tomato Cultivation

Abstract

In greenhouse tomato cultivation, three primary methods of flower pollination exist: insect pollination, physical pollination by vibrating flowers, and artificial pollination using hormone-based chemicals. Insect pollination, the natural method, involves insects (e.g., honeybees) vibrating flowers to collect pollen and nectar. This paper proposes an alternate approach, using small drones to search and pollinate flowers in place of bees autonomously. We report field experiments conducted using these drone technologies. The drone must locate flowers ready for pollination. We developed an artificial intelligence (AI) image classification system (AI classifier) using machine learning to identify these flowers. Equipped with an AI classifier, the drone searches for flowers through autonomous flight and positioning technology. Upon identifying a suitable flower during its search, the drone makes contact to pollinate it. Integrating AI-based flower detection, autonomous flight control for flower search, and a pollination control device allows the drone to perform pollination. This study devises these technologies, implements them in a drone, and evaluates their effectiveness through a pollination experiment.

Keywords

• Autonomous drones
• greenhouse tomato cultivation
• pollination methods
• pollination
• AI image classification
• flower search