见叶知果:迈向自动化苹果幼果疏除
机器人学
2023-02-21 v1 计算机视觉与模式识别
摘要
顺应全球趋势,可靠技能劳动力的匮乏正对苹果园的有效管理造成严峻问题。主要挑战之一是维持能够做出精确幼果疏除决策的熟练人力。疏除需要准确测量单棵苹果树的实际负载,以逐树提供最优疏除决策。这是一项艰巨任务,因为茂密枝叶遮挡了树体结构中的幼果。本文介绍了一台自动化苹果幼果疏除机器人视觉系统的初步设计、实现与评估细节,以满足上述需求。该平台由 UR5 机械臂与立体相机组成,使其能绕开叶片观测,从而映射苹果枝条上幼果的精确数量与大小。我们表明,该平台在真实商业苹果园中能以 84% 的准确率与 87% 的精度测量树体幼果负载。
引用
@article{arxiv.2302.09716,
title = {Seeing the Fruit for the Leaves: Towards Automated Apple Fruitlet Thinning},
author = {Ans Qureshi and Neville Loh and Young Min Kwon and David Smith and Trevor Gee and Oliver Bachelor and Josh McCulloch and Mahla Nejati and JongYoon Lim and Richard Green and Ho Seok Ahn and Bruce MacDonald and Henry Williams},
journal= {arXiv preprint arXiv:2302.09716},
year = {2023}
}
备注
Accepted and Presented at the Australasian Conference on Robotics and Automation (ACRA 2022)