温室智能循迹小车系统稳定性研究  

Study on the stability of greenhouse intelligent tracking car system

作　　者：陈璇 张立勇 陈超 孙鹏 魏芳坤 张收港 CHEN Xuan;ZHANG Liyong;CHEN Chao;SUN Peng;WEI Fangkun;ZHANG Shougang(School of Mechanical Engineering,Anhui Institute of Science and Technology,Chuzhou 233100,China;Weisong Optoelectronics Technology Co.,Ltd.,Hefei 230088,China;Headfree Technology Co.,Ltd.,Hefei 230094,China)

机构地区：[1]安徽科技学院机械工程学院,安徽滁州233100 [2]安徽唯嵩光电科技有限公司,安徽合肥230088 [3]安徽明生恒卓科技有限公司,安徽合肥230094

出　　处：《电子设计工程》2023年第18期90-94,99,共6页Electronic Design Engineering

基　　金：安徽省教育厅协同创新项目(GXXT-2019-020);安徽科技学院科技支撑项目(2021zrzd0l)。

摘　　要：基于新时期智慧农业无人管理的实际需求,该文系统以NVIDIA Jetson Nano为主板,充分使用OpenCV数据库函数,并设计了多色彩路径实验方法对小车进行测试,得出了小车在静态、动态下目标定位的准确率,通过调整无线通信模块及室内的网络条件,使静态准确率由86%提升至100%。同时,还对小车进行了卷积神经网络模型的训练学习,测试了小车在特征颜色路径下的循迹执行情况。试验及验证结果表明,小车整体识别准确率较高且循迹稳定性较好,具备在智慧农业温室推广使用的条件。Based on the actual demand of unmanned management of intelligent agriculture in the new era,this system uses NVIDIA Jetson Nano as the main board,makes full use of OpenCV database function,and designs multi⁃color path experiment method to test the car,and obtains the accuracy of the car’s target positioning under static and dynamic conditions.By adjusting the wireless communication module and indoor network conditions,the static accuracy rate is improved from 86%to 100%.At the same time,the convolutional neural network model of the car is trained and studied,and the tracking performance of the car under the feature color path is tested.The experimental and verification results show that,the overall identification accuracy and tracking stability of the vehicle are high,and the vehicle has the conditions for promoting and using in the smart agricultural greenhouse.

关 键 词：循迹 OpenCV视觉库 温室 Jetson Nano 

分 类 号：TN919.82[电子电信—通信与信息系统]