Simulation Study on Tokamak Relevant Visual Servoing System

Tokamak is a plasma confinement device that can be used to achieve magnetically confined nuclear fusion within a reactor. Owing to the harsh environment, human access is not possible inside the reactor and Remote Handling (RH) systems are used for inspection and maintenance of the tokamak in-vessel components. These RH systems are long and cantilevered articulated robotic arms. Thus, inherent structural deformations or minor errors in assembly cause the position of the Tool Center Point (TCP) to vary largely than the anticipated position by the joint controller. Visual Servoing can rectify these errors by providing a photometric information to the control system using feedback from a pre-calibrated camera mounted on the arm of the robot. This paper presents the development and simulation of a visual-servo mechanism for a 6-axis tokamak relevant Articulated Robotic Inspection Arm (ARIA). The image based visual servo controller for ARIA is implemented using a C++ based controller. The controller subscribes the in-vessel component images from a camera topic and computes the instantaneous joint velocity based on difference in errors of the image feature. A virtual reality simulation is performed to show the response of ARIA based on the controller output. A short review of visual servoing is presented along with the kinematics of the ARIA arm. Results from the simulations show a good convergence of thejoint trajectories and smooth alignment of the TCP with respect to the in-vessel component.