Delivery of pollen to forsythia flower pistils autonomously and precisely using a robot arm

Abstract

Autonomous pollination robots have received significant attention in recent years. It is expected that these robots will assist farmers in pollinating plants effectively. Despite employing various pollination methods, delivering pollen autonomously and precisely to pistils (the reproductive organs of flowers) remains a challenge. This is due to the small size of pistils, as well as difficulty in their rotation and orientation identification. Under this background, we present a novel method for detecting the orientation and area of the rotated pistil, known as PSTL_Orient, which we combine with servoing techniques and flower detection to develop an autonomous pollination robot. The experimental results demonstrate the superiority of our proposed PSTL_Orient in comparison to existing state-of-the-art methods in the detection of the orientation and rotation of pistil anchors. Additionally, the proposed pollination robot can achieve a 91.08 % accuracy in detecting flowers and an 86.19 % success rate in pollinating the small pistils of forsythia flowers. By being mounted on a ground vehicle, the proposed robotic arm system can help farmers in practical large-scale greenhouse environments.