基于 YOLOv10s 和外接球法融合的粮食收获机卸粮口位姿检测方法研究

Research on the grain harvester unloading outlet shape and position detection based on YOLOv10s and circumsphere fusion

摘要

[目的]目前粮食收获机田间作业时存在卸粮口位姿检测精度低的问题，导致收获机与无人运粮车“机-车”协同卸运作业效率低下，本文旨在提出一种基于 YOLOv10s 和外接球法的粮食收获机卸粮口位姿检测方法。[方法]使用 YOLOv10s 目标检测技术、Canny 边缘分割和外接球法，实现对粮食收获机卸粮口的质心定位和开口姿态精准检测。同目标检测直接定位算法和平均质心法开展检测性能对比分析，并分析其在不同光照环境下的适应性。[结果]本研究算法在 3000lx 以上和 100~300lx 环境下 5m 内定位误差分别小于 1.20%和 0.99%，开口姿态角度误差分别小于 8.8°和 7.3°。相较目标检测直接定位算法和平均质心法，3000lx 环境质心定位检测精度分别提高 11.20%和 3.51%，100~ 300lx 环境质心定位检测精度分别提高 10.92%和 4.06%。定位精度显著提升，并实现较低的开口方向检测误差。[结论]本研究提出的粮食收获机卸粮口位姿检测技术，能精准检测粮食收获机卸粮口的位姿信息，可为精准判断卸粮口出粮落点，以及后续根据运粮车仓内粮食分布进行路径优化提供基础数据研究。

Abstract

[Objectives] Currently, there is a low accuracy in the unloading port pose detection of grain harvesters during field operations, leading to low efficiency in the "machine-vehicle" collaborative unloading operations between harvesters and unmanned grain transport vehicles. This study proposed a grain harvester unloading port pose detection method based on YOLOv10s and the minimum bounding sphere method. [Methods] It utilized the YOLOv10s object detection technology, Canny edge segmentation, and the minimum bounding sphere method to achieve precise detection of the centroid location and the opening pose of the harvester unloading port. The performance of this method was compared with direct localization algorithms for object detection and the average centroid method, and its adaptability in different lighting environments was also studied. [Results] The proposed algorithm achieved positioning errors of less than 1.20% and 0.99% within 5 meters in environments with above 3000 lx and environments with 100-300 lx, respectively. The opening angle errors were less than 8. 8° and 7. 3°, respectively. Compared to target detection-based positioning algorithms and centroid positioning methods, the centroid positioning accuracy was improved by 11.20% and 3.51% in strong light environments, and by 10.92% and 4.06% in weak light environments. The positioning accuracy was significantly improved, with a relatively low opening direction detection error achieved. [Conclusions] The grain harvester unloading port pose detection technology proposed in this study could accurately detect the pose information of the unloading port, providing fundamental data for accurately determining the unloading point and subsequent path optimization based on the grain distribution in the transport vehicle's storage.

关键词

目标检测 / 边缘分割 / 卸粮口 / 位姿 / 外接球 / YOLOv10s

Key words

target detection / edge segmentation / unloading port / centroid localization / pose detection / circumsphere / YOLOv10s

参考文献

[1] 罗锡文, 胡炼, 何杰, 等. 中国大田无人农场关键技术研究与建设实践[J]. 农业工程学报 2024, 40(1): 1-16.

[2] Luo X W, Hu L, He J, et al. Key technologies and practice of unmanned farm in China[J]. Transactions of the Chinese Society of Agricultural Engineering, 2024, 40(1): 1-16(in Chinese with English abstract).

[3] 张漫, 季宇寒, 李世超, 等. 农业机械导航技术研究进展[J]. 农业机械学报, 2020, 51(4): 1-18.

[4] Zhang M, Ji Y H, Li S C, et al. Research progress of agricultural machinery navigation technology[J]. Transactions of the Chinese Society for Agricultural Machinery, 2020, 51(4): 1-18(in Chinese with English abstract).

[5] 郑和, 李紫琪, 单以才, 等. 农田机械除草技术与装备研究现状与发展趋势[J]. 南京农业大学学报, 2025, 48(2): 285-295. DOI: 10.7685/jnau.202404014.

[6] Zheng H, Li Z Q, Shan Y C, et al. Current status and development trends of agricultural machinery weeding technologies and equipment[J]. Journal of Nanjing Agricultural University, 2025, 48(2): 285-295(in Chinese with English abstract).

[7] 满忠贤, 何杰, 刘善琪, 等. 智能农机多机协同收获作业控制方法与试验[J]. 农业工程学报, 2024, 40(1): 17-26.

[8] Man Z X, He J, Liu S Q, et al. Method and test for operating multi-machine cooperative harvesting in intelligent agricultural machinery[J]. Transactions of the Chinese Society of Agricultural Engineering, 2024, 40(1): 17-26(in Chinese with English abstract).

[9] 李文鑫, 张璠, 姚竟发, 等. 联合收割机与运粮车协同作业调度技术研究[J]. 中国农机化学报, 2023, 44(2): 119-125.

[10] Li W X, Zhang F, Yao J F, et al. Research on cooperative operation scheduling technology of combine harvester and grain truck[J]. Journal of Chinese Agricultural Mechanization, 2023, 44(2): 119-125(in Chinese).

[11] 李文鑫. 联合收割机与运粮车协同作业调度与避障技术研究[D]. 保定:河北农业大学, 2023.

[12] Li W X. Research on cooperative operation scheduling and obstacle avoidance technology of combine harvester and grain truck[D]. Baoding:Hebei Agricultural University, 2023(in Chinese with English abstract).

[13] 张昊. 联合收获机自动卸粮系统的设计与试验[D]. 镇江:江苏大学, 2023.

[14] Zhang H. Design and testing of an automatic grain unloading system for combine harvesters[D]. Zhenjiang:Jiangsu University, 2023(in Chinese with English abstract).

[15] 张闻宇, 张智刚, 张帆, 等. 水稻收获转运双机协同自主作业策略与试验[J]. 农业工程学报, 2022, 38(15): 1-9.

[16] Zhang W Y, Zhang Z G, Zhang F, et al. Cooperative autonomous operation strategy and experiment of the rice harvester together with a rice-transporting vehicle[J]. Transactions of the Chinese Society of Agricultural Engineering, 2022, 38(15): 1-9(in Chinese with English abstract).

[17] 张闻宇, 张智刚, 罗锡文, 等. 收获机与运粮车纵向相对位置位速耦合协同控制方法与试验[J]. 农业工程学报, 2021, 37(9): 1-11.

[18] Zhang W Y, Zhang Z G, Luo X W, et al. Position-velocity coupling control method and experiments for longitudinal relative position of harvester and grain truck[J]. Transactions of the Chinese Society of Agricultural Engineering, 2021, 37(9): 1-11(in Chinese with English abstract).

[19] Zhang W Y, Hu L W, Ding F, et al. Parking precise alignment control and cotransporter system for rice harvester and transporter[J]. Computers and Electronics in Agriculture, 2023, 215: 108443.

[20] Liu Z P, Dhamankar S, Evans J T, et al. Automation of agricultural grain unloading-on-the-go[J]. IFAC-Papers on Line, 2022, 55(24): 248-253.

[21] 卢心缘, 赵华民, 任锐, 等. 基于 YOLOv5s-seg 的非结构化枣园视觉导航路径提取[J]. 南京农业大学学报, 2024, 47(6): 1241-1250. DOI: 10.7685/jnau.202312032.

[22] Lu X Y, Zhao H M, Ren R, et al. Unstructured jujube garden visual navigation path extraction based on YOLOv5s-seg[J]. Journal of Nanjing Agricultural University, 2024, 47(6): 1241-1250(in Chinese with English abstract).

[23] Pini M, Marucco G, Falco G, et al. Experimental test bed and methodology for the assessment of RTK GNSS receivers used in precision agriculture[J]. IEEE Access, 2020, 8: 14690-14703.

[24] 李培艺, 汪小旵, 王延鑫, 等. 基于 LiDAR 多维点云优化的垄作菊花采摘机器人自主导航方法研究[J]. 南京农业大学学报, 2024, 47(4): 809-822. DOI: 10.7685/jnau.202308019.

[25] Li P Y, Wang X C, Wang Y X, et al. Autonomous navigation system for chrysanthemum picking robot based on LiDAR multidimensional point cloud optimization[J]. Journal of Nanjing Agricultural University, 2024, 47(4): 809-822(in Chinese with English abstract).

[26] 刘超, 陈锦明, 刘慧, 等. 基于三维激光雷达与优化 DBSCAN 算法的果树定位方法[J]. 农业机械学报, 2023, 54(4): 214-221,240.

[27] Liu C, Chen J M, Liu H, et al. Fruit tree location method based on 3D LiDAR and optimized DBSCAN algorithm[J]. Transactions of the Chinese Society for Agricultural Machinery, 2023, 54(4): 214-221,240(in Chinese with English abstract).

[28] 徐文静, 冼锦洪, 孙东松. 用于烟尘监测的偏振激光雷达系统及实验研究[J]. 红外与激光工程, 2023, 52(3): 195-204.

[29] Xu W J, Xian J H, Sun D S. Polarization lidar system for smoke monitoring and experimental research[J]. Infrared and Laser Engineering, 2023, 52(3): 195-204(in Chinese).

[30] 范安琪, 左常洲, 彭菁, 等. 基于双目立体视觉的冻干草莓外部品质等级检测方法研究[J]. 南京农业大学学报, 2023, 46(2): 396-404. DOI: 10.7685/jnau.202201010.

[31] Fan A Q, Zuo C Z, Peng J, et al. Study on detecting method of external quality level of freeze-dried strawberry based on binocular stereo vision[J]. Journal of Nanjing Agricultural University, 2023, 46(2): 396-404(in Chinese with English abstract).

[32] Zhang W, Huang H, Sun Y Q, et al. AgriPest-YOLO:a rapid light-trap agricultural pest detection method based on deep learning[J]. Frontiers in Plant Science, 2022, 13: 1079384.

[33] Zhao H Y, Tang Z Z, Li Z H, et al. Real-time object detection and robotic manipulation for agriculture using a YOLO-based learning approach[C]// 2024 IEEE International Conference on Industrial Technology(ICIT). March 25-27, 2024,Bristol,United Kingdom. IEEE, 2024: 1-6.

[34] Bosilj P, Aptoula E, Duckett T, et al. Transfer learning between crop types for semantic segmentation of crops versus weeds in precision agriculture[J]. Journal of Field Robotics, 2020, 37(1): 7-19.

[35] Su D, Kong H, Qiao Y L, et al. Data augmentation for deep learning based semantic segmentation and crop-weed classification in agricultural robotics[J]. Computers and Electronics in Agriculture, 2021, 190: 106418.

基金资助

国家重点研发计划项目(2022YFD2001604)