体育动作多媒体数据库在采摘机器人执行器上的应用  

The Application of Sports Action Multimedia Database in Picking Robot Actuator

作　　者：苏献军 Su Xianjun(Guangxi Police College, Nanning 530023,China)

机构地区：[1]广西警察学院,南宁530023

出　　处：《农机化研究》2022年第6期236-239,244,共5页Journal of Agricultural Mechanization Research

基　　金：广西高校中青年教师科研基础能力提升项目(2020KY0139)。

摘　　要：采摘机器人动作的规范性和准确性不仅关系到果实的采摘效率,还会影响到采摘的效果,特别是易损果实,很容易造成果实的破碎,降低采摘质量。为了提高采摘机器人动作的准确性,基于体育训练项目的多媒体数据库,结合神经网络机器学习训练和图像处理技术,对采摘机器人动作进行了训练和优化,并对动作的准确性进行了仿真模拟。结果表明:优化后的采摘动作误差较小,且随着神经网络训练样本的增加,误差有下降趋势,为采摘机器人采摘末端的设计提供了重要的依据。The standardization and accuracy of the action of the picking robot are not only related to the picking efficiency of the fruit,but also affect the picking effect,especially the vulnerable fruit,which is easy to cause the breakage of the fruit and reduce the picking quality.In order to improve the accuracy of picking robot action,it trained and optimized the picking robot action,and simulated the accuracy of the action based on the multi-media database of sports training items,combined with neural network machine learning training and image processing technology.The simulation results show that the optimized picking action error is small,and with the neural network training With the increase of samples,the error has a downward trend,which provides an important basis for the design of the picking end of the picking robot.

关 键 词：采摘机器人 多媒体数据库 动作优化 神经网络 图像处理 

分 类 号：S225[农业科学—农业机械化工程] TP242[农业科学—农业工程]