This study introduces a robotic pollination approach designed to address the decline of natural pollinators and enhance agricultural productivity. The system integrates a mobile robotic platform, a collaborative robotic arm, and an RGB-D camera for vision-based flower detection. It autonomously navigates to target plants, identifies fully bloomed flowers, and executes pollination through precise end-effector control. Motion planning is achieved using an Incremental Roadmap (IRM)-based algorithm, which generates collision-free trajectories to avoid damaging delicate flowers and surrounding plant structures. The platform was tested in an indoor setting, where it successfully demonstrated flower localization and effective pollination within a controlled environment. These results confirm the feasibility of using autonomous robotic systems for pollination tasks, providing a consistent and efficient alternative to manual or natural pollination methods.

While the system achieved promising results in controlled conditions, adapting it for real-world agricultural environments remains a key challenge. Future work will focus on enhancing robustness under variable lighting, irregular plant geometries, and dynamic conditions that are typically encountered in greenhouses or open fields. Improvements in vision algorithms, real-time environmental perception, and adaptive motion planning will be essential for ensuring operational reliability. Additionally, the integration of advanced machine learning techniques could further optimize flower detection accuracy and pollination efficiency across diverse crop types.

Overall, this research demonstrates the potential of robotic pollination systems as a scalable, sustainable solution for modern agriculture. By combining advanced sensing, motion planning, and AI-driven decision-making, these systems could significantly reduce labor dependency, improve pollination consistency, and contribute to higher crop yields.