Investigation of safety mechanisms for oceanographic AUV missions programming

• Published in: Europe Oceans 2005 Vol. 2; pp. 906 - 913 Vol. 2
• Main Authors: Nana, L., Marce, L., Opderbecke, J., Pettier, M., Rigaud, V.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 2005
• Subjects: Artificial intelligence; Functional analysis; Intelligent sensors; Intelligent systems; Intelligent vehicles; Logic; Marine safety; Sea surface; Sensor systems; State feedback
• ISBN: 9780780391031; 0780391039
• DOI: 10.1109/OCEANSE.2005.1513177

Our investigation of the safety of AUV applications concerns a global architecture for mission preparation, supervision and execution and the functional analysis of the architecture from the dependability point of view. At the mission preparation level, a safety subsystem is proposed. This subsystem takes as input the actions/tasks of the AUV represented into a generic format, and an informal description of the mission, and generates firstly, with the help of the scientist, a formal description of the mission, as well as constraints extracted from the informal description of the mission. These outputs are then used together with the vehicle parameters and the possible specific properties provided by the scientist, to perform verifications like satisfaction of constraints and properties such as the interparameters adequacy of actions of the mission and satisfaction of mission logics in comparison with pre- and post-conditions of actions. At the surface supervision level, in addition to the possibilities provided to the operator through classical information feedback coming from the vehicle (sensors data, execution state of actions and failures state), an intelligent diagnosis system, based on the IFREMER ADVOCATE II model, makes it possible to perform a more deepened diagnosis on the basis of vehicle data and of diagnosis information synthesis coming from the embedded intelligent diagnosis module. The vehicle control subsystem level uses, similarly to the surface supervision level, intelligent diagnosis modules based on different artificial intelligence techniques.