Autonomous Mobile Manipulations for Valve Turning


Student thesis: Master's Thesis


Robots are one of the first thoughts that comes to mind when imaging the future. They are anticipated to play a major role in different aspects of life, and perform tasks more efficiently and accurately. To achieve such a futuristic goal, robots should be capable of perceiving the surrounding environment, and reacting autonomously. This project is inspired by Mohamed Bin Zayed International Robotics Challenge (MBZIRC). To be more specific, it is a contribution towards the second challenge, where an unmanned ground vehicle is expected to navigate its way to a panel, detect a set of wrenches hanged on this panel, guide the robotic arm towards them, and finally grasp a wrench to rotate a valve stem. Since the second challenge is very broad, it is divided into the described stages. The work provided in this thesis mainly targets the problem of autonomouslyguiding the robot manipulator to the best pre-grasping pose, with aid of visual servoing. Achieving such tasks would increase the performance of robots in disaster response applications, such as oil spills. This thesis provides a thorough literature review detailing the main challenges and milestones in visual servoing, with a discussion on basic vision based control techniques and their advantages/disadvantages. Then it presents the designing steps of classical visual servoing tasks with simulation results in both Matlab and Gazebo. In addition, it includes hardware implementation of the Image Based Visual Servoing approach. Finally, it describes alternative solutions to classical methods, such as the new Wavelets-based family. Indexing Terms: Visual servoing, multi-resolution wavelets transform, visual tracking, autonomous arm manipulation, MBZIRC.
Date of AwardDec 2016
Original languageAmerican English
SupervisorJorge Dias (Supervisor)


  • Visual servoing
  • multi-resolution wavelets transform
  • visual tracking
  • autonomous arm manipulation

Cite this