Abstract: A 3D vision-guided robot flexible motion platform was constructed, and control software based on the ROS 2 architecture was developed. The basic functions of the platform were tested, achieving robot flexible motion guided by 3D vision. In this platform, 3D reconstruction and point cloud registration are realized through the collaboration of a binocular structured light camera and a turntable. Point, line, and surface information of the workpiece is identified using the Random Sample Consensus Algorithm. High-precision motion paths for the robot are planned according to preset tasks, and online motion control is achieved via the TCP/IP protocol. The platform can be used for experimental teaching in mechanical and electrical majors, conducting comprehensive experiments on robot 3D vision coordinate system calibration. It also serves as a research and development platform for 3D vision algorithms and robot path planning algorithms.