莲花
莲花效应
工程类
机制(生物学)
人工智能
计算机科学
模拟
植物
生物
生态学
认识论
哲学
原材料
作者
Yuqi Zeng,Long Xue,Boshi Tang,Chaoyang Ying,Muhua Liu,Jing Li,Keke Liao,Chengzhi Ruan
摘要
ABSTRACT Addressing the issue of the absence of relevant automated harvesting machinery for lotus seedpods, which currently relies solely on manual harvesting, this paper proposes a four‐degree‐of‐freedom cantilever‐type lotus seedpod harvesting device. This harvesting device comprises an electric mobile chassis, mobile mechanism, harvesting mechanism, transfer mechanism, image acquisition system, and control system, which make it suitable for harvesting lotus seedpod in standardized lotus fields. The cantilevered lotus seedpod picking device obtains the image of the lotus from the depth camera at an overlooking angle. The YOLOv5‐trained lotus‐bud recognition model is used to identify and locate the lotus seedpod. Through the calibration using active vision method, the conversion relationship between the ZED camera coordinates and the base coordinates of the picking device is calculated to realize lotus seedpod picking. The results of picking experiments conducted in different time periods show that the picking success rate is higher at night, and in the early morning and evening, with the picking success rates of 89.47%, 85.7%, and 85%, respectively, while the picking success rate at noon is only 42.86%. It can be seen that the ambient light has a great influence on the picking of the lotus seedpods, and it is reasonable to avoid the midday period with strong light for picking. The experimental results show that the clamping coupling method realizes the integration of automatic picking and conveying of lotus seedpods, improves the picking efficiency, and provides a new scheme for fruit picking.
科研通智能强力驱动
Strongly Powered by AbleSci AI