Research and Application of Object Recognition Method of Visual Grasping Robot Based on Deep Learning

In order to improve the flexibility and intelligence of the grasping robots in the field of logistics, a research on visual robot objects recognition method based on deep learning was carried out. Aiming at this problem, a real-time object recognition and grasp detection scheme is proposed, the rectangle of the items to be operated is defined, the convolutional neural network structure model is established, and the loss function is defined. Finally, experimental verification was performed, including data processing and network model training, and the results analysis showed the effectiveness of this research method.