Image-based visual servoing for autonomous crossing of narrow passages in hostile environments

At European Organization for Nuclear Research (CERN), measurement, inspection and maintenance tasks are a daily occurrence. In order to achieve job automation and to minimize relative work risks, custom made robotic platforms are gradually replacing human operators. The MIRA SPS Robot, an omnidirectional manipulator-equipped mobile platform, is the one used at the Super Proton Synchrotron particle accelerator. This thesis is about the automation of the narrow passages crossing, that occurs every time the robot enters a new sector of the accelerator. The proposed strategy uses a visual control algorithm based on detection and tracking of known geometric features. Validation has been done at first on simulation, and later in the real case scenario.