| 基于双目视觉的香蕉园巡检机器人导航路径提取方法 |
| |
| 引用本文： | 张振乾,李世超,李晨阳,曹如月,张漫,李寒,李修华.基于双目视觉的香蕉园巡检机器人导航路径提取方法[J].农业工程学报,2021,37(21):9-15. |
| |
| 作者姓名： | 张振乾 李世超 李晨阳 曹如月 张漫 李寒 李修华 |
| |
| 作者单位： | 1. 中国农业大学现代精细农业系统集成研究教育部重点实验室，北京 100083;2. 中国农业大学农业农村部农业信息获取技术重点实验室，北京 100083;3. 广西大学电气工程学院，南宁 530004 |
| |
| 基金项目： | 广西创新驱动发展专项资金（桂科AA18118037）；国家重点研发计划（Grant No. 2019YFB1312300-2019YFB1312305）；中国农业大学建设世界一流大学（学科）和特色发展引导专项资金（2021AC006） |
| |
| 摘 要： | 为实现移动机器人香蕉园巡检自动导航，研究提出了一种基于双目视觉的香蕉园巡检路径提取方法。首先由机器人搭载的双目相机获取机器人前方点云，进行预处理后对点云感兴趣区域进行二维投影并将投影结果网格化，得到网格地图；然后采用改进的K-means算法将道路两侧香蕉树分离，其中初始聚类中心通过对网格地图进行垂直、水平投影以及一、二阶高斯拟合确定；最后基于最小包围矩形提取导航路径，将道路两侧网格以最小矩形框包围，提取两包围框中间线作为期望导航路径。测试结果表明，改进的K-means算法聚类成功率为93%，较传统方法提高了12%；导航路径提取平均横向偏差为14.27 cm，平均航向偏差为4.83°，研究方法可为香蕉园巡检机器人自动导航提供支持。
|
| 关 键 词： | 机器人 机器视觉 香蕉园 巡检 路径提取 |
| 收稿时间： | 2021/5/15 0:00:00 |
| 修稿时间： | 2021/11/18 0:00:00 |
Navigation path detection method for a banana orchard inspection robot based on binocular vision |
| |
| Zhang Zhenqian,Li Shichao,Li Chenyang,Cao Ruyue,Zhang Man,Li Han,Li Xiuhua.Navigation path detection method for a banana orchard inspection robot based on binocular vision[J].Transactions of the Chinese Society of Agricultural Engineering,2021,37(21):9-15. |
| |
| Authors: | Zhang Zhenqian Li Shichao Li Chenyang Cao Ruyue Zhang Man Li Han Li Xiuhua |
| |
| Institution: | 1. Key Laboratory of Modern Precision Agriculture System Integration Research, Ministry of Education, China Agricultural University, Beijing 100083, China;2. Key Laboratory of Agricultural Information Acquisition Technology, Ministry of Agriculture and Rural Affairs,China Agricultural University, Beijing 100083, China;; 3. School of Electrical Engineering, Guangxi University, Nanning 530004, China |
| |
| Abstract: | Abstract: An anti-drift sprayer with a conical wind field has emerged as an innovative structural device for the auxiliary airflow in plant protection operations during crop production. There is also a significant reduction of droplet loss for the effective deposition of fine particles in the target areas. However, an airflow obstruction can be found in the flow channel of the current sprayers, leading to the lower overall performance of the device. In this study, a systematic optimization was made on the inner flow channel in an anti-drift spray device under a conical wind field, thereby clarifying the anti-drift mechanism for better performance of the device. A force model of a single droplet was also constructed for the moving air medium, according to proton dynamics. After that, the CFD numerical simulation and wind field test were utilized to optimize the airflow loss in the inner runner for the better design of the device. The simulation result showed that the disturbance of eddy current was improved without the abnormality of local speed after optimization. Specifically, the conical wind speed at the outlet of the device still reached 17.00 m/s, increasing by 23.5%, compared with that before the optimization. The wind speed test showed that the effective utilization rate of the auxiliary airflow at the outlet of the device was 21.2% higher than that before the optimization, when the radius of curvature of the inner flow channel elbow was designed to be 4 cm, indicating that the optimization plan was feasible. Furthermore, there were significant correlations between the cross wind speed, nozzle height, conical wind speed, and the proportion of total droplet drift under wind tunnel conditions. By contrast, there was a negative correlation between the conical wind speed and the proportion of total droplet drift. More importantly, the proportion of the total droplet drift presented a downward trend, whereas, the adverse effect of the crosswind on the droplet deposition gradually decreased, with the increase of conical wind speed. Additionally, a multi-factor orthogonal experiment was carried out to establish the mathematical model of the total droplet drift ratio in the vertical/horizontal direction. It was found that the cone-shaped wind field significantly reduced the droplet loss in the space with the wind. There was also a higher significance of the vertical/horizontal mathematical model (P<0.05, R2=0.934/0.945). Consequently, the conical wind field can be widely expected to effectively resist the generation of vortexes, thereby reducing the droplet loss with the wind in the vertical height. This finding can also provide a sound reference for the comprehensive analysis of droplet migration and deposition in protected agriculture. |
| |
| Keywords: | robots machine vision banana orchard inspection path extraction |
|
| 点击此处可从《农业工程学报》浏览原始摘要信息 |
| 点击此处可从《农业工程学报》下载免费的PDF全文 |
|
