Visual Servo

For my university image processing class, I created a fun project for my self learning image processig programming. The goal was to develop a system that uses computer vision to control and guide a robot or device based on visual feedback. I implemented basic image processing techniques to detect and track the poisition of the red ball in real time. The control of Z1 manipulator arm is handled by Moveit2. The below video is conducted in Gazebo simulation.

The core method involved using color segmentation and contour detection to identify the target object in the camera feed. Once detected, the system calculated the object’s position and sent commands to adjust the manipulator arm movement, keeping the target centered in view. The movement of Go2 (quadruped robot) was manually controlled by joystick. This hands-on project helped me understand practical applications of image processing and control systems, especially the usage of OpenCV and Moveit2. The video below showing the Z1’s hand camera perspective.