A laboratory platform for teaching computer vision control of robotic systems

Abstract

Nowadays, due to the involved complex issues, teaching robotics is demanding new methodologies. One of the complex issues regarding robotics is the use of computer vision for a proper robot guiding control. At the University of Brasilia an educational robotic workstation around a Rhino XR4 robot was built. Students can perform experiences on kinematics, trajectory generation and motion control. A new controller was further designed to introduce motion control by visual servoing. These new laboratory experiments allow implementation of various vision-based control schemes: position-based (3D), image-based (2D) and hybrid (21/2D). For visual servoing simulation experiments a robot wire-frame modeling scheme simulation RobSim was developed within the Matlab environment. The simulation environment is flexible regarding parameter adjustments for robot, camera, targets and workspace. As the simulation experiment becomes satisfactory, the experiment can then be performed with the robotic workstation. Primitive functions built on Matlab allow modeling and simulating a large class of manipulators.