Method for identifying and positioning workpiece key points by robot

The invention provides a method for identifying and positioning key points of a workpiece by a robot. According to the invention, a direct mapping relationship from a workpiece physical graph to a robot operation point is established; a key point position of construction is determined in a photographing mode; training is carried out through a deep convolutional neural network to obtain a key pointrecognition model; the key points, namely teaching points, on the workpiece can still be accurately positioned when the workpiece is zoomed, translated, rotated and slightly deformed and distorted, so that the identification of the key points of the workpiece is not influenced by the placement position, the rotation angle and the tiny deformation of the workpiece, the key point identification error is small, and the manufacturing cost is low. 提供一种机器人对工件关键点识别与定位的方法，本发明建立了工件实物图到机器人作业点直接的映射关系，通过拍照的方式确定施工的关键点位，通过深度卷积神经网络进行训练得到关键点识别模型，能够在工件位置发生缩放、平移、旋转、微小形变扭曲时，仍然能够准确定位工件上关键点即示教点，使工件关键点位的识别不受工件摆放位置、旋转角度、微小形变的影响，关键点识别的误差较小，