Multi-Sensor Data Fusion for a Long-Reach Robot

Abstract: The purpose of this Master's thesis project is the study and the improvement of the positioning of a long-reach robot. The AIA robot is a 10 meter long multi-link manipulator developed at CEALIST (France), and designed for close inspection and maintenance of the inner walls of a thermonuclear reactor (tokamak) under vacuum and high temperature conditions. Previous tests campains pointed out the insu-ciency of its positioning due to an inaccurate joint angle sensor measure. In order to estimate precisely the angular position of the robot's joints, a reconstruction strategy using both motor and joint sensors measurements is considered. The reconstruction is achieved with the help of a Kalman lter based on the joint transmission model. The Hammerstein model structure with an hysteresis nonlinearity is proposed to describe the behaviour of the transmission between the motor and the joint. After o-line validation using Matlab, the resulting reconstruction procedure was implemented in C in the robot's real-time controller.