English

中国农机化学报

中国农机化学报 ›› 2024, Vol. 45 ›› Issue (7): 34-40.DOI: 10.13733/j.jcam.issn.2095-5553.2024.07.006

• 农业装备工程 • 上一篇    下一篇

果蔬采摘动作采集系统设计与试验

肖莹1,谢铭露2,景一佳3,童一飞3   

  1. 1. 北京电子科技职业学院,北京市,100176; 2. 农业农村部南京农业机械化研究所,南京市,210014;
    3. 南京理工大学机械工程学院,南京市,210094

  • 出版日期:2024-07-15 发布日期:2024-06-21
  • 基金资助:
    中央高校基本科研业务费专项资金资助项目(30919011205)

Design and experiment of fruit and vegetable picking action collection system

Xiao Ying1, Xie Minglu2, Jing Yijia3, Tong Yifei3   

  1. 1. Beijing Polytechnic, Beijing, 100176, China; 2. Nanjing Institute of Agricultural Mechanization,
    Ministry of Agriculture and Rural Affairs, Nanjing, 210014, China; 3. School of Mechanical Engineering,
    Nanjing University of Science and Technology, Nanjing, 210094, China
  • Online:2024-07-15 Published:2024-06-21

摘要: 为提高果蔬采摘机器人动作的准确性,建立一套可应用于农作物采摘的动作数据库。基于视觉识别技术,结合农业生产中对于果蔬采摘动作判断的实际需求进行分析研究,分别从硬软件设计的角度,设计一种针对果蔬采摘动作的综合采集系统。在硬件层面,重点介绍包括UPS电源、运算主机、摄像头等各部件的需求分析及选型结果。在软件层面,介绍软件系统总体架构并将子功能模块化处理,此外重点介绍各功能模块间的软硬件交互方式。最后以苹果的采摘动作为研究对象,设计整机测试试验及动作识别试验来检验系统的完备性。试验结果表明,该动作采集系统对于单一采摘动作的识别具有较高的正确率,识别率达100%。

关键词: 果蔬采摘, 动作采集, 视觉识别, 动作数据库

Abstract: In order to improve the accuracy of action of fruit and vegetable picking robot, it is established a set of action database which can be applied to crop picking. Based on visual recognition technology, this paper analyzed and studied the actual needs of fruit and vegetable picking action judgment in agricultural production, and designed a comprehensive collection system for fruit and vegetable picking action from the perspective of hardware and software design. On the hardware level, the paper mainly introduced the demand analysis and selection results of various components including UPS power supply, computing host and camera. At the software level, it introduced the overall architecture of the software system and modularized the subfunctions. In addition, it focused on the interaction between the software and hardware among the functional modules. Finally, taking the apple picking action as the research object, the whole machine test and the action recognition test were designed to check the completeness of the system. The test results showed that the action acquisition system had a high accuracy for single picking action recognition, the recognition rate was 100%.

Key words: fruit and vegetable picking, action collection, visual recognition, action database

中图分类号: