English

中国农机化学报

中国农机化学报 ›› 2024, Vol. 45 ›› Issue (9): 202-208.DOI: 10.13733/j.jcam.issn.2095-5553.2024.09.031

• 农业信息化工程 • 上一篇    下一篇

一种轮式采花机器人设计——以黄蜀葵为例

桑一男 1,徐增莱 2,汪琼 2,葛海涛 3,王殿广 3   

  1. (1. SYD Dynamics ApS,Denmark,Odense M,5230;2.江苏省中国科学院植物研究所(南京中山植物园),江苏省植物资源研究与利用重点实验室,南京市,210014;3.江苏苏中药业研究院有限公司,南京市,210018)
  • 出版日期:2024-09-15 发布日期:2024-09-04
  • 基金资助:
    南京市科技计划(国际联合研发项目)(202002008)

Design of a wheeled flower picking robot: Taking Abelmoschus manihot (Linn.) Medicus as an example 

Sang Yinan1,Xu Zenglai2,Wang Qiong2,Ge Haitao3,Wang Dianguang3   

  1. (1. SYD Dynamics ApS,Odense M,5230,Denmark;2. Jiangsu Key Laboratory for the Research and Utilization of Plant Resources,Institute of Botany,Jiangsu Province and Chinese Academy of Sciences(Nanjing Botanical Garden Mem. Sun Yat.Sen),Nanjing,210014,China; 3. Jiangsu Suzhong Pharmaceutical Research Institute Co.,Ltd.,Nanjing,210018,China) 
  • Online:2024-09-15 Published:2024-09-04

摘要:

摘要:针对黄蜀葵花朵特定采收的时间和人工采摘效率过低的问题,为实现各类花朵采摘机械化和智能化的需求,以黄蜀葵花朵为例并结合其生长特性,设计一种轮式采花机器人。采用工控机和嵌入式微控制器作为采花机器人的主控系统,执行机构采用电力驱动,以蓄电池供电,通过多个推杆电机、舵机和动态基座(基于航姿参考系统,可自动调节工作平台倾角的机构)构成轮式行走机构,采用两个摄像头分别同时获取黄蜀葵花朵图像,通过深度识别算法以筛选识别可采摘的目标,采用多个舵机配合带轮结构构成机械臂和夹爪的控制机构,以完成花朵的采摘与收集。试验结果表明,机器人对花朵定位的准确率可达 75%,识别率高达 100%。采花机器人通过主控制系统在试验田里能成功完成采摘作业,由机械臂配合其夹爪成功抓取花朵,上位机软件可以完成图像采集识别、机械臂控制和机器人工作路线的行驶等操作。该机器人适用于各地黄蜀葵和其他部分植物花朵的采集。

关键词: 黄秋葵, 机器视觉, 深度学习, 农业自动化, 花朵采摘, 惯性导航系统

Abstract:

In order to solve the problem that the specific harvesting time and the manual picking efficiency of Abelmoschus manihot are too low,and to meet the needs of mechanization and intelligence in picking all kinds of flowers,a wheeled flower picking robot is designed by taking Abelmoschus manihot(Linn.) Medicus as an example and combining their growth characteristics. The system adopts industrial computer and embedded microcontroller as the main control system of the machine. The actuator of the machine is powered by electric drive in the form of battery. The wheel type walking mechanism is composed of multiple pusher motor actuators and dynamic bases(a mechanism that can automatically adjust the inclination of the working platform based on the attitude reference system). Two cameras are used to obtain the flower images of hollyhock respectively, and the depth recognition algorithm is used to screen and identify the picking targets. Multiple steering gears and pulley structure are used to form the control structure of mechanical arm and gripper to complete the picking and collection of flowers. The experimental results show that the robot can locate flowers with 75% accuracy and 100% recognition rate. The picking robot can successfully complete the picking operation in the test field through the main control system. The robot arm cooperates with its gripper to successfully grasp the flowers. The upper computer software can complete the operations such as image acquisition and recognition,robot arm control and robot working route driving. It is suitable for the collection of yellow marshmallow and other plant flowers. 

Key words: Abelmoschus manihot, machine vision, deep learning, agricultural automation, flower picking, inertial navigation system ,

中图分类号: