Eye-in-hand visual servoing with a robotic arm

Abstract

One of the major roles of robotic manipulators is pick and place tasks in automated assembly, hence by adding real-time visual feedback, robotic manipulators can ac-complish autonomous tasks. The aim of this project is to design, implement and test an eye-in-hand robotic arm system that is able to position its end-effector above a static object in the workspace.

In this dissertation, a USB camera was attached to the gripper of a 5 degree of freedom (DOF) robotic manipulator. The static object was chosen to be a square with three of its corners marked in red. The image frames were then processed using a MATLAB script file. The chosen image processing technique was colour recognition, which detects the three highlighted corners of the target and returns the centroid position in pixels of these corners. A velocity control law was then used to process these results and obtain the necessary camera velocities to reach the target. Finally the velocities are sent to a Simulink model containing a velocity controller which controls the actuation of the robotic arm. The effectiveness of the implemented controller has been illustrated and verified through a series of testing methodology.