Precision spraying using variable time delays and vision-based velocity estimation

被引:3
作者
Sanchez, Paolo Rommel [1 ,2 ]
Zhang, Hong [2 ]
机构
[1] Univ Philippines, Coll Engn & Agroind Technol, Inst Agr & Biosyst Engn, Agribiosyst Machinery & Power Engn Div,Agrometeoro, Los Banos 4031, Laguna, Philippines
[2] Rowan Univ, Henry M Rowan Coll Engn, Mech Engn Dept, Glassboro, NJ 08028 USA
来源
SMART AGRICULTURAL TECHNOLOGY | 2023年 / 5卷
关键词
Weed; Spraying; Modular robot; Precision agriculture; Site-specific agriculture; Convolutional neural networks; Machine vision; Computer vision; MACHINE VISION; WEED-CONTROL; DESIGN; SYSTEM;
D O I
10.1016/j.atech.2023.100253
中图分类号
S2 [农业工程];
学科分类号
0828 ;
摘要
Traditionally, precision farm equipment often relies on real-time kinematics and global positioning systems (RTK-GPS) for accurate position and velocity estimates. This approach proved effective and widely adopted in developed regions where RTK-GPS satellite and base station availability and visibility are not limited. However, RTK-GPS signal can be limited in farm areas due to topographic and economic constraints. Thus, this study developed a precision sprayer that estimated the travel velocity locally by tracking the relative motion of plants using a deep-learning-based machine vision system. Sprayer valves were then controlled by variable time delay (VTD) queuing and dynamic filtering. The proposed velocity estimation approach was tested at different velocities and tracking thresholds. The results showed that the velocity estimates agreed with actual measurements with a mean absolute error of 0.036 m/s. Further, testing the targeting algorithm on rows of artificial crops and weeds at different levels of spraying duration and filter size factor (FSF) showed that short spraying duration and small FSF increase overall spraying accuracy. Finally, testing the sprayer using the optimum settings at 0.87 m/s and 1.03 m/s successfully sprayed all targets. Further, only 2% to 7% of non-targets were sprayed at the low and high test velocities, respectively. With these results, this study suggests that vision-based velocity estimation combined with VTD queuing and dynamic filtering can be an accurate and low-cost solution for targeted spraying without using auxiliary velocity measurement systems.
引用
收藏
页数:13
相关论文
共 50 条
[31]   A machine vision-based method for crowd density estimation and evacuation simulation [J].
Huang, Shijie ;
Ji, Jingwei ;
Wang, Yu ;
Li, Wenju ;
Zheng, Yuechuan .
SAFETY SCIENCE, 2023, 167
[32]   Vision-based force measurement using neural networks for biological cell microinjection [J].
Karimirad, Fatemeh ;
Chauhan, Sunita ;
Shirinzadeh, Bijan .
JOURNAL OF BIOMECHANICS, 2014, 47 (05) :1157-1163
[33]   Vision-based housing price estimation using interior, exterior & satellite images [J].
Nouriani, Ali ;
Lemke, Lance .
INTELLIGENT SYSTEMS WITH APPLICATIONS, 2022, 14
[34]   Real-Time (Vision-Based) Road Sign Recognition Using an Artificial Neural Network [J].
Islam, Kh Tohidul ;
Raj, Ram Gopal .
SENSORS, 2017, 17 (04)
[35]   Vision-based excavator pose estimation for automatic control [J].
Liu, Guangxu ;
Wang, Qingfeng ;
Wang, Tao ;
Li, Bingcheng ;
Xi, Xiangshuo .
AUTOMATION IN CONSTRUCTION, 2024, 157
[36]   A Method of Vision-based State Estimation of an Unmanned Helicopter [J].
Yuan, Bin ;
Hao, Yingguang .
2010 6TH INTERNATIONAL CONFERENCE ON WIRELESS COMMUNICATIONS NETWORKING AND MOBILE COMPUTING (WICOM), 2010,
[37]   A software platform for vision-based UAV autonomous landing guidance based on markers estimation [J].
Xu XiaoBin ;
Wang Zhao ;
Deng YiMin .
SCIENCE CHINA-TECHNOLOGICAL SCIENCES, 2019, 62 (10) :1825-1836
[38]   Fast vision-based pose estimation iterative algorithm [J].
Li, Long ;
Deng, Zong-Quan ;
Li, Bing ;
Wu, Xiang .
OPTIK, 2013, 124 (12) :1116-1121
[39]   A fusion framework for vision-based indoor occupancy estimation [J].
Sun, Kailai ;
Liu, Peng ;
Xing, Tian ;
Zhao, Qianchuan ;
Wang, Xinwei .
BUILDING AND ENVIRONMENT, 2022, 225
[40]   Vision-Based Surface Inspection System for Bearing Rollers Using Convolutional Neural Networks [J].
Wen, Shengping ;
Chen, Zhihong ;
Li, Chaoxian .
APPLIED SCIENCES-BASEL, 2018, 8 (12)