欢迎您访问《智慧农业(中英文)》官方网站! English

Smart Agriculture ›› 2022, Vol. 4 ›› Issue (3): 63-74.doi: 10.12133/j.smartag.SA202207008

• 专刊--智慧果园关键技术与装备 • 上一篇    下一篇

果园自主导航兼自动对靶喷雾机器人

刘理民1,2(), 何雄奎1,2(), 刘伟洪3, 刘紫嫣1,2, 韩虎1,2, 李扬帆1,2   

  1. 1.中国农业大学 理学院,北京 100193
    2.中国农业大学 农业无人机系统研究院,北京 100193
    3.中国农业大学 工学院,北京 100083
  • 收稿日期:2022-07-16 出版日期:2022-09-30 发布日期:2022-11-04
  • 基金资助:
    国家梨产业体系(CARS-28);国家自然科学基金资助项目(31761133019);中国农业大学2115人才培育发展支持计划;三亚中国农业大学研究院引导资金项目(SYND-2021-06)
  • 作者简介:刘理民(1994-),男,博士研究生,研究方向为智慧果园及智能植保机械。E-mail:liulimsy2882@163.com
  • 通讯作者: 何雄奎 E-mail:liulimsy2882@163.com;xiongkui@ cau.edu.cn

Autonomous Navigation and Automatic Target Spraying Robot for Orchards

LIU Limin1,2(), HE Xiongkui1,2(), LIU Weihong3, LIU Ziyan1,2, HAN Hu1,2, LI Yangfan1,2   

  1. 1.College of Science, China Agricultural University, Beijing 100193, China
    2.College of Agricultural Unmanned System, China Agricultural University, Beijing 100193, China
    3.College of Engineering, China Agricultural University, Beijing 100083, China
  • Received:2022-07-16 Online:2022-09-30 Published:2022-11-04
  • corresponding author: HE Xiongkui E-mail:liulimsy2882@163.com;xiongkui@ cau.edu.cn

摘要:

为同时实现果园智能植保机自主导航及自动对靶喷雾,研制了一种果园自主导航兼自动对靶喷雾机器人。首先采用单个3D LiDAR(Light Detection and Ranging)采集果树信息确定兴趣区(Region of Interest,ROI),对ROI内点云进行2D化处理得到果树质心坐标,通过随机一致性(Random Sample Consensus,RANSAC)算法得到果树行线,并确定果树行中间线(导航线),进而控制机器人沿导航线行驶。通过编码器及惯性测量单元(Inertial Measurement Unit,IMU)确定机体速度及位置,IMU矫正采集到的果树分区冠层信息,最后通过程序判断分区冠层的有无控制喷头是否喷雾。结果表明,机器人自主导航时最大横向定位偏差为21.8 cm,最大航向偏角为4.02°,相比于传统连续喷雾机施药液量、空中漂移量及地面流失量分别减少20.06%、38.68%及51.40%。本研究通过单个3D LiDAR、编码器及IMU在保证喷雾效果的前提下,实现了喷雾机器人自主导航及自动对靶喷雾,降低了农药使用量及飘失量。

关键词: 自主导航, 对靶喷雾, LiDAR, 随机一致性算法, 机器人, 惯性测量单元

Abstract:

To realize the autonomous navigation and automatic target spraying of intelligent plant protect machinery in orchard, in this study, an autonomous navigation and automatic target spraying robot for orchards was developed. Firstly, a single 3D light detection and ranging (LiDAR) was used to collect fruit trees and other information around the robot. The region of interest (ROI) was determined using information on the fruit trees in the orchard (plant spacing, plant height, and row spacing), as well as the fundamental LiDAR parameters. Additionally, it must be ensured that LiDAR was used to detect the canopy information of a whole fruit tree in the ROI. Secondly, the point clouds within the ROI was two-dimension processing to obtain the fruit tree center of mass coordinates. The coordinate was the location of the fruit trees. Based on the location of the fruit trees, the row lines of fruit tree were obtained by random sample consensus (RANSAC) algorithm. The center line (navigation line) of the fruit tree row within ROI was obtained through the fruit tree row lines. The robot was controlled to drive along the center line by the angular velocity signal transmitted from the computer. Next, the ATRS's body speed and position were determined by encoders and the inertial measurement unit (IMU). And the collected fruit tree zoned canopy information was corrected by IMU. The presence or absence of fruit tree zoned canopy was judged by the logical algorithm designed. Finally, the nozzles were controlled to spray or not according to the presence or absence of corresponding zoned canopy. The conclusions were obtained. The maximum lateral deviation of the robot during autonomous navigation was 21.8 cm, and the maximum course deviation angle was 4.02°. Compared with traditional spraying, the automatic target spraying designed in this study reduced pesticide volume, air drift and ground loss by 20.06%, 38.68% and 51.40%, respectively. There was no significant difference between the automatic target spraying and the traditional spraying in terms of the percentage of air drift. In terms of the percentage of ground loss, automatic target spraying had 43% at the bottom of the test fruit trees and 29% and 28% at the middle of the test fruit trees and the left and right neighboring fruit trees. But in traditional spraying, the percentage of ground loss was, in that sequence, 25%, 38%, and 37%. The robot developted can realize autonomous navigation while ensuring the spraying effect, reducing the pesticides volume and loss.

Key words: automatic navigation, automatic target spraying, LiDAR, random sample consensus algorithm, robot, IMU

中图分类号: