采摘机器人目标识别分类方法研究——基于LM优化算法和BP神经网络

Research on Target Recognition and Classification of Picking Robot--Based on LM Optimization Algorithm and BP Neural Network

为使采摘机器人更加快速和准确地识别目标和进行等级分类,提出了一种基于LM优化算法的BP神经网络的目标识别分类方法,能够快速对目标水果进行识别和分类。测试结果表明:采摘机器人目标识别分类方法对测试样本的分类正确率在90%以上,图像识别时间在3.80s以内,能够满足采摘机器人作业需求,证实了该方法的可行性。

In order to make the picking robot recognize and classify the target more quickly and accurately,it proposes a method of target recognition and classification based on BP neural network of LM optimization algorithm,which can recognize and classify the target fruit quickly.The test results show that the accuracy of target recognition and classification method to test samples is more than 90%,and the time of image recognition is less than 3.80s,which can meet the needs of robot operation so as to prove the feasibility of the method.