基于深度学习的摄像头与机械臂融合视觉感知与目标识别算法研究

Research on the Fusion of Camera and Robot Arm Visual Perception and Object Recognition Algorithm Based on Deep Learning

摘要: 目前，随着深度学习、机器视觉以及机械臂控制技术的迭代更新，使用深度学习对目标物体进行检测、识别、控制以发展成为新兴研究领域。本文主要研究一种基于深度学习的摄像头与机械臂视觉感知与目标识别算法。通过研究深度学习算法和ROS2操作系统的基本原理和应用，针对机械臂、摄像头视觉感知与目标识别典型应用场景，提出了先进行相机标定，保证摄像头采集图像的成像质量和精度，采用YOLOv8目标检测算法实现对目标物体的准确识别，最后使用机械臂PID控制算法实现机械臂的精确运动控制抓取、移动目标物。通过反复实验，表明该算法能够实现对目标物体的快速精准识别，并通过机械臂实现目标物体精确运动控制。

Abstract: Currently, with the iterative updates of deep learning, machine vision, and robotic arm control technologies, the use of deep learning for detecting, recognizing, and controlling target objects has become an emerging research field. This article mainly studies a deep learning based visual perception and object recognition algorithm for cameras and robotic arms. By studying the basic principles and applications of deep learning algorithms and ROS2 operating system, this paper proposes to first calibrate the camera to ensure the imaging quality and accuracy of the captured images for typical application scenarios of robotic arms, camera visual perception, and object recognition. The YOLOv8 object detection algorithm is used to achieve accurate recognition of the target object, and finally the PID control algorithm of the robotic arm is used to achieve precise motion control of the robotic arm for grasping and moving the target object. Through repeated experiments, it has been shown that the algorithm can achieve fast and accurate recognition of target objects, and achieve precise motion control of target objects through a robotic arm.