Intelligent Collaborative Robot Based on DMHF-CNN+YOLOv7 and Soft Q-Learning in Non-Standard Industrial Parts Assembly

The aim of this study is to explore a novel intelligent collaborative robot system that incorporates DMHF-CNN+YOLOv7 target detection technology, hybrid continuous-discrete reinforcement learning motion planning, and Soft Q-Learning-based intelligent grasping control technology. Experimental results show that the system exhibits good performance and stability in non-standard industrial parts assembly tasks. The accuracy of target detection reaches more than 95 percent, the accuracy of motion planning and the success rate of grasping are significantly improved, and the assembly time and the cost of human intervention are effectively reduced. The research results in this paper provide strong support for the application of intelligent manufacturing technology in actual production.