Research on an Autonomous Recognition and Grasping Manipulator System Based on YOLOv8

This paper presents an intelligent robotic arm system leveraging YOLOv8 to tackle perception delays and low target recognition accuracy in complex battlefield environments. By integrating six-axis robotic arm kinematics with deep-learning visual detection, we develop a cohesive "perception-decision-execution" framework. Initially, target detection utilizes the YOLOv8 model, with spatial pose derived from binocular vision depth calculations, enhanced by training on a specialized dataset. Subsequently, a kinematic model of the robotic arm is created, with MATLAB simulations analyzing the end-effector's workspace and enabling autonomous trajectory planning. Finally, a Python-based control system facilitates autonomous recognition and grasping. Experimental results demonstrate high target recognition accuracy and grasping success rates in complex settings, offering a viable solution for intelligent equipment development in military and industrial applications.