基于改进YOLOv4-tiny的球形果蔬定向系统设计  

Spherical Fruit and Vegetable Orientation System Based on Improved YOLOv4-tiny

作　　者：黄铭 张宏伟[1,2] 王振东 赵文帆 Huang Ming;Zhang Hongwei;Wang Zhendong;Zhao Wenfan(School of Electrical Engineering and Automation,Henan Polytechnic University,Jiaozuo,Henan 454000,China;Henan Key Laboratory of Intelligent Detection and Control of Coal Mine Equipment,Jiaozuo,Henan 454003,China)

机构地区：[1]河南理工大学电气工程与自动化学院,河南焦作454000 [2]河南省煤矿装备智能检测与控制重点实验室,河南焦作454003

出　　处：《机电工程技术》2024年第2期132-136,192,共6页Mechanical & Electrical Engineering Technology

基　　金：河南省大学生创新创业训练计划项目(202210460056);河南省高等教育教学改革研究与实践项目(2021SJGLX1011);河南省高校科技创新团队项目(20IRTSTHN019)。

摘　　要：球形果蔬的高效定向是其深加工过程中的关键环节。针对传统定向装置存在的结构复杂、成本高、效率低等问题,设计了一种球形果蔬自动定向系统。对其系统控制方案进行了设计,并提出了一种基于目标检测技术的位姿调整方法。针对定向机构上苹果偏转角度较大时果梗检测困难的问题,以YOLOv4-tiny模型为基础,通过引入CBAM注意力机制来增强网络表征能力,提升了果梗检测算法的性能,降低了漏检率。为了改善形状不规则果实的定向效果,设计了一种位姿调整模糊控制器,提高算法的鲁棒性。搭建位姿调整模拟试验台进行定向实验。结果表明:定向的合格率为90%,平均时间为3.1 s,平均误差为4.6 mm,计算量低,实时性好,为自动定向的实际应用提供可行的方法。The high efficiency orientation of spherical fruit and vegetable is the key link in its deep processing.Aiming at the problems of complex structure,high cost and low efficiency of traditional orientation device,an automatic orientation mechanism is designed.Its system control scheme is designed,and a pose adjustment method based on object detection technology is proposed.Aiming at the difficulty of calyx detection when the deflection angle of apple is large in orientation mechanism,based on YOLOv4-tiny model,CBAM attention mechanism is introduced to enhance the network characterization ability,thus improving the performance of calyx detection algorithm and reducing the missed detection rate.In order to improve the orientation of irregularly shaped fruit,a fuzzy controller for position adjustment is designed to improve the robustness of the algorithm.The orientation experiment is carried out on the pose adjustment simulation testbed.The results show that the percent of pass is 90%,the average orientation time is 3.1 s,the average error is 4.6 mm,the calculation amount is low,and the realtime performance is good,which provides a feasible method for the practical application of automatic orientation.

关 键 词：果蔬定向 目标检测 注意力机制 模糊控制 

分 类 号：TP271[自动化与计算机技术—检测技术与自动化装置]