A stereoscopic eye-in-hand vision system for remote handling in ITER

The International Thermonuclear Experimental Reactor (ITER) maintenance is performed by means of remote handling (RH) systems and with aid of user interfaces such as haptic and joystick devices, virtual reality (VR) systems, and camera views. Many RH operations involving RH equipment, such as robotic manipulator arms, require millimeter accuracy, but camera views are often occluded or of poor quality, and might be unavailable during sensitive steps that require accurate, close-up views. Moreover, the VR system may not reflect the current scene accurately, as physical conditions may have changed under the harsh environment. The purpose of this research was to prototype and evaluate a novel software system, called 3D Node, that locates and detects the position and orientation of a piece of RH equipment or reactor element with respect to a stereo camera pair. The detection information is utilized to adjust the motion trajectories of a robotic manipulator arm. The 3D Node features stereo-camera calibration, target depth mapping, target position and orientation detection, and online target tracking. This paper reports on the 3D Node demonstration on the ITER Divertor RH use case and discusses the system applicability to other ITER RH systems.