In Dehua pear cultivation, manual picking is inefficient, with labor costs exceeding 15% per mu. Traditional mechanical picking, due to the rigidity of its equipment, often results in a fruit breakage rate exceeding 20%, making it difficult to adapt to the thin skin (0.2-0.3 mm) and crispy flesh of the pear. To address this industry pain point, this paper designs an integrated harvesting robot system using a ‘bionic end effector+YOLOv11 visual positioning’ approach. The core of the system consists of a six-degree-of-freedom robotic arm, a binocular depth camera, and an electrically driven, separate end effector. The actuator utilizes a three-finger flexible gripping and shearing mechanism, balancing non-destructive grasping with precise stalk severing. The visual system, based on YOLOv11, introduces the C2PSA attention module to enhance the distinction between fruit and leaf features, and combines it with a binocular camera for three-dimensional positioning. Experiments based on samples from a pear orchard in Dehua, Fujian, show that the recall rate for the ‘pear’ category remained above 0.85 at a confidence level≥0.7, with an optimal F1 value of 0.83 (confidence level 0.565) and a stable mAP50 of 0.87. Field tests also demonstrate that the system achieved four times the efficiency of manual picking, while keeping the fruit breakage rate below 5%. This solution provides technical support for automated Dehua pear harvesting, and its design principles are valuable for the development of harvesting equipment for fragile fruits such as peaches and strawberries.
为获取德化梨果柄的空间位置信息,本研究采用训练完成的目标检测模型及对应权重文件,搭配Basler ace-acA2500-14gc双目深度相机构建定位流程:图像采集阶段,通过相机软件开发工具包(software development kit,SDK)调用RGB(red,green,blue)与深度流并行工作模式,确保每一帧彩色图像与对应深度数据同步获取,避免因数据异步导致的定位偏差;针对模型检测输出的果柄二维边界框,提取框中心像素坐标作为关键定位点,结合双目相机测得的该像素点深度值,代入相机内参、主点坐标,通过反投影公式将二维图像点转化为相机坐标系下的三维坐标。该方法可精准获取果柄空间位置,为后续末端执行器抓取姿态调整、机械臂避障路径规划提供可靠的空间数据支撑,适配德化梨果柄剪切的精准作业需求。相机坐标系转换关系示意图及工作流程如图10所示,图10中:Xw、Yw、Zw为世界坐标系,描述目标在真实三维空间中的位置;Xc、Yc、Zc为相机坐标系,以相机光心为原点建立的三维坐标系; R 为旋转矩阵; T 为平移向量;[ R │ T ]为外参矩阵; K 为相机内矩阵区;Pworld为空间点在世界坐标系下的齐次坐标;为空间点在图像坐标系下的齐次坐标;为空间点在相机坐标系下的三维坐标;u,v为图像像素坐标系的坐标轴。
ZHENGQ S.Current status and countermeasures of pear industry development in Dehua[J].South China Agriculture,2022,16(1):63-66.
[3]
KARKEEM, ADHIKARIB, AMATYAS,et al.Identification of pruning branches in tall spindle apple trees for automated pruning[J].Computers and Electronics in Agriculture,2014,103:127-135.
[4]
PORTOLESI P.Toward sustainable harvests:Designing an automated picking robot for peaches[J/OL].2024.Available at:
[5]
LAMMERSK, ZHANGK X, ZHUK Y,et al.Development and evaluation of a dual-arm robotic apple harvesting system[J].Computers and Electronics in Agriculture,2024,227:109586.
[6]
SILWALA, DAVIDSONJ R, KARKEEM,et al.Design,integration,and field evaluation of a robotic apple harvester[J].Journal of Field Robotics,2017,34(6):1140-1159.
[7]
QADDOORI FENJANS, FATHOLLAHI DEHKORDIS.Soft robotic system with continuum manipulator and compliant gripper:Design,fabrication,and implementation[J].Actuators,2024,13(8):298.
[8]
HUSSAINM, HEL, SCHUPPJ,et al.Green fruit removal dynamics for development of robotic green fruit thinning end-effector[J].Journal of the ASABE,2022,65(4):779-788.
[9]
HOHIMERC J, WANGH, BHUSALS,et al.Design and field evaluation of a robotic apple harvesting system with a 3D-printed soft-robotic end-effector[J].Transactions of the ASABE,2019,62(2):405-414.
[10]
DAVIDSONJ R, MOC.Mechanical design and initial performance testing of an apple-picking end-effector[C]//Proceedings of IMECE 2015,ASME 2015 International Mechanical Engineering Congress and Exposition, November 13-19,2015.Houston,Texas,USA.American Society of Mechanical Engineers,2015,57397:V04 AT 04A011.
[11]
PARKW, SEOS,BAE J.A hybrid gripper with soft material and rigid structures[J].IEEE Robotics and Automation Letters,2019,4(1):65-72.
[12]
庞国友.手动背负夹爪式水果(梨)采摘器设计与试验研究[D].长沙:中南林业科技大学,2020.
[13]
PANGG Y.Design and experimental study of a manual back-carrying claw-type fruit (pear) picker[D].Changsha:Central South University of Forestry and Technology,2020.
TENGC, DONGX B, SONGZ K,et al.Target detection and localization of pinecone based on improved YOLOv5s[J].Forest Engineering,2025,41(4):812-826.
[16]
FANP, YANB, WANGM R,et al.Three-finger grasp planning and experimental analysis of picking patterns for robotic apple harvesting[J].Computers and Electronics in Agriculture,2021,188:106353.
[17]
HUG R, CHENC, CHENJ,et al.Simplified 4-DOF manipulator for rapid robotic apple harvesting[J].Computers and Electronics in Agriculture,2022,199:107177.
[18]
CHAIJ F, CHENY, ZHANGY Y,et al.System research of pear picking robot[C]//Proceedings of the 2022 4th International Conference on Robotics,Intelligent Control and Artificial Intelligence.Dongguan China.ACM,2022:107-112.
[19]
LIM, LIUP.A bionic adaptive end-effector with rope-driven fingers for pear fruit harvesting[J].Computers and Electronics in Agriculture,2023,211:107952.
[20]
GAOJ, ZHANGF, ZHANGJ X,et al.Development and evaluation of a pneumatic finger-like end-effector for cherry tomato harvesting robot in greenhouse[J].Computers and Electronics in Agriculture,2022,197:106879.
ZHANGY Y, TIANJ Q, WANGW X,et al.Design and experiments of end-effector for dragon fruit picking robot[J].Journal of Forestry Engineering,2023,8(4):144-150.
[23]
孟飞.圣女果采摘机械臂关键技术的研究[D].天津:天津农学院,2023.
[24]
MENGF.Research on the key technology of the picking manipulator of the virgin fruit[D].Tianjin:Tianjin Agricultural University,2023.
ZHANGR H, OUJ J, DINGL X,et al.Research on series-parallel hybrid passion fruit picking robot based on vision positioning[J].Journal of Chinese Agricultural Mechanization,2023,44(10):209-216,244.
ZHUL X, LAIY J, ZHANGS A,et al.Improved U-Net dragon pitaya image segmentation and pose estimation method for picking robot[J/OL].Transactions of the Chinese Society for Agricultural Machinery,2023:1-16.[2025-06-15]
LIUF, LIUY K, LINS,et al.Fast recognition method for tomatoes under complex environments based on improved YOLO[J].Transactions of the Chinese Society for Agricultural Machinery,2020,51(6):229-237.
[35]
王登奎.基于深度学习的苹果识别及采摘点定位系统研究[D].秦皇岛:燕山大学,2023.
[36]
WANGD K.Research on apple recognition and picking point location system based on deep learning[D].Qinhuangdao:Yanshan University,2023.
XIH Y, ZHANGD, ZHOUT,et al.Research on the method of picking up overlapping tomato fruits by picking robot[J].Journal of Agricultural Mechanization Research,2021,43(12):17-23,50.
ZHANGQ H, KONGD X, LIJ Q,et al.Design of fruit picking system based on inverse kinematics dimension reduction and YOLO v4[J].Transactions of the Chinese Society for Agricultural Machinery,2021,52(9):15-23.
KODORSS, SONDORSM, LĀCISG,et al.Rapid prototyping of pear detection neural network with YOLO architecture in photographs[J].Environment Technology Resources Proceedings of the International Scientific and Practical Conference,2023,1:81-85.