Intelligent robot grabbing method based on 3D vision

The invention relates to an intelligent robot grabbing method based on 3D vision. The intelligent robot grabbing method comprises the steps: 1, completing hand-eye calibration of an intelligent grabbing system; 2, completing the training of a grabbing planning model based on a convolutional neural network; and 3, completing grabbing of the intelligent robot in a real environment. The method has the advantages thatby means of the intelligent robot autonomous grabbing method based on 3D vision, 3D object recognition and positioning, autonomous collision-free path planning and other work can be completed in an unstructured environment, the method has very important application value in industrial robot picking, sorting, stacking, assembling and other work environments, and the self-adaptive capacity of the industrial robot under the characteristics of various 3C industry types and rapid production line updating can be remarkably improved. 本发明涉及一种基于3D视觉的智能机器人抓取方法，包括：步骤1，完成智能抓取系统的手眼标定；步骤2，完成基于卷积神经网络的抓取规划模型的训练；步骤3，