Hoang Anh PHAM (Oct. 2017)
Funding : Ministry
Supervision : Thierry SORIANO
Autonomous Underwater Vehicles are increasingly used for oceans scientific surveys. They aim at operating in groups.
The choice of an architecture scheme of generic supervision control that allows to generate a set of cooperative vehicles is of primal importance to ensure inter-operability of the whole system.
In a decentralized scheme, distributed control leads to a good upgradability and a certain robustness. If a failure occurs on one of the vehicles, it should be still possible to reassign the mission on the remaining swarm, which increases the robustness with respect to he mission. Drawbacks are to be evaluated accurately ; available computer power of each vehicle can be a limiting factor to solve complex problems; The information available from other vehicles can be wrong, incomplete or delayed, which can create a risk of collision that is non negligible in case of malfunction. The work proposed in the context of this thesis will explore in detail the modeling and the quantification of these performance and robustness aspects. The design of cooperative control algorithms requires to take into account the behavior of a vehicle with the behaviors of the others. In order to be more efficient, the control laws should be based on the current and planned states of all the vehicles. The global behavior of such systems is therefore complex and will be addressed thanks to hybrid formalism. It is planned to develop detailed behavioral scenarios and cooperative control algorithms devoted to coordination for monitoring missions, which will be first simulated then implemented.
Reuse, modularity and specialization are all to be associated with the production of a new application, and constitute the strong points of the object-oriented approaches. It will be necessary to write down the customizing and reusing rules of the cooperative control elements that were recently developed. Actually these are points that were studied in the case of a single autonomous underwater vehicle with an object-oriented methodology, which we would like to reuse and extend to a cooperation of autonomous underwater vehicles in the context of system engineering and opensource environment.