Application of deep learning in target grasping of machine arm

Inhalt:
Deep learning method can be applied in many advanced fields to improve the performance and result. For example, in the area of machine arms, deep learning algorithms can be applied in the process of object detection and trajectory planning. This paper focuses on what kind of deep learning methods can be used in machine arms and how they can improve the performance of robot arms when grasping. In object detection process, multiple methods like Yolo, SSD and transformer can be used. As the technique of object detection is very mature, robot arms have high accuracy classifying their target. In the process of trajectory planning, methods like RetinaNet and SE-RetinaGrasp can be used. These algorithms are able to find out the proper places which have the highest possibility for grasping tasks. The application of these methods has greatly improved the accuracy and intelligence of robot arms. Nowadays, machine arms can perform many tasks which used to be handled only with the help of human. In the future, more research can be done in improving trajectory planning methods so that the algorithm can perform in high speed and accuracy while finding the proper place to hold on its target when there are multiple objects in the background.