Research on Intelligent Assembly Strategy and Workpiece Grasping Method for Industrial Robots Based on Deep Learning

Abstract

In response to the current situation of low assembly accuracy and unreasonable workpiece grasping posture in the automatic assembly process of equipment manufacturing based on industrial robots, an objective function was designed with the goal of minimizing robot grasping torque, and a deep learning strategy was used to autonomously identify the optimal grasping posture. In terms of assembly strategy selection, the assembly behavior is abstracted as the coordination between holes and shafts. A method of changing the center distance of shaft hole parts to change the jamming state of holes and shafts is proposed to increase the assembly qualification rate. Finally, the industrial robot in the training base is used as the experimental object to validate the method proposed in this paper. After comparative analysis, the proposed method increases the assembly efficiency by 10.4%, and the assembly success rate reaches 96%.