Systeme de navigation inertielle par filtrage de Kalman indirect pour un sous-marin d'inspection robotis

Conventional positioning techniques using an inertial navigation system are known to drift with time. They offer poor performance for applications requiring high precision such as underwater remotely operated vehicle (ROV). This thesis proposes to use an inertial navigation system and merge data with external measurement to improve orientations and positions stability. This technique will allow the Hydro-Quebec Research Institute (IREQ) underwater inspection vehicule to better reference anomalies contain on dams surface.;The main objective of the research project is to integrate by indirect Kalman filter measures of an inertial system with an absolute underwater positioning system to estimate the position of an ROV. To achieve this goal, the inertial navigation system will be analyzed and studied. The algorithms will be first tested in simulation before being implemented in a real system. The performance of the data fusion will be validated by imposing real underwater vehicle movement to the inertial sensor. A robotic manipulator will move the sensors according to an ROV trajectory. This approach will provide a good repeatability. The excellent accuracy of the actual ROV position will facilitates the evaluation of the algorithms performance.;Experimental results as show that the use of an indirect Kalman filtering as a solution to the positioning problem of an underwater inspection vehicule is valid. However, some constraints must be respected so that the performance of the algorithm is satisfactory.