智能系统学报2023，Vol.18Issue(6)：1259-1267,9.DOI:10.11992/tis.202212026
基于语义分割视觉伺服的种苗自动夹取系统设计
Design of automatic picking system for seedlings based on semantic segmentation visual servo
摘要
Abstract
Modern plant tissue culture is a time-consuming and labor-intensive task with monotonous work.An automat-ic seedling clamping system based on semantic segmentation visual servo was designed and tested to reduce labor costs and increase production.First,a vision localization method was proposed based on the DP-BiSeNetV2 semantic seg-mentation algorithm to determine the appropriate clamping point on the root.Further,a clamping device suitable for the actual working environment was designed,developed,and tested.Finally,an automatic seedling clamping system was constructed by integrating the vision localization algorithm with the robot clamping device.In the experimental session,tests were conducted using the Phalaenopsis seedling dataset.In the semantic segmentation experiment,the mIoU and pixel accuracy of the DP-BiSeNetV2 model were 63.51%and 98.25%,respectively.Furthermore,the success rate was 81.7%in the clamping experiment.Experimental results show that the automatic clamping system has a large potential to meet the transplantation requirements of plant tissue culture production lines.关键词
语义分割/视觉伺服/种苗/夹取点定位/夹取系统/机械手/智能抓取/深度学习Key words
semantic segmentation/visual servo/seedling/grasping point localization/grasping system/robot arm/intel-ligent robotic grasping/deep learning分类
信息技术与安全科学引用本文复制引用
翟永杰,王家豪,张鑫,胡东阳,王乾铭,徐大伟,刘亚军..基于语义分割视觉伺服的种苗自动夹取系统设计[J].智能系统学报,2023,18(6):1259-1267,9.基金项目
国家自然科学基金联合基金重点支持项目(U21A20486) （U21A20486）
中国科学院自动化研究所复杂系统管理与控制国家重点实验室开放课题(20220102). （20220102）
