In order to improve the accuracy and speed of grasping industrial parts by visual control manipulator in industrial production,a new detection algorithm for identifying industrial parts categories and optimal grasping position was proposed.YOLOv5l target detection algorithm was used to identify a variety of industrial parts in the field of vision,and then the identified pictures were transferred into the grab position detection algorithm to identify the best grasping position.Aiming at the problem of grasping position detection,an improved neural network model was proposed.On the basis of GG-CNN network,a four-layer residual network was added to do flat layer feature extraction and enhance the effect of feature extraction.The results show that the recognition accuracy of this algorithm is above 95%,and the success rate of grasping is about 90%.It is verified that the algorithm has high accuracy and timeliness in the recognition of various industrial parts and the best grasping position.