Decision making in multi-UAVs systems: Architecture and Algorithms

International audienceThis chapter depicts an architecture that aims at designing a multi-UAV framework enabling cooperative operations in a system in which some UAVs are directly controlled by an operator, others are only endowed with operational autonomy , and others have decisional autonomy capacities. The architecture provides with the possibility to configure dynamically the decisional scheme, depending on the available robots and on the operational context. A taxonomy of robots decisional autonomy is introduced, and used as a foundation to state the proposed architecture. The various functionalities on-board each robot are organized among a repartition that exhibits on-board functional components , and on-board or on-ground generic executive and decision making processes. A set of algorithms that fulfill the three main decision-making functionalities required in a multi-robot system are then presented: a contract-net protocol that can handle task allocation for complex multi-UAV missions, a planning scheme based on a Hierarchical Task Networks planner completed with plan-refiners that consider the actual domain models, and an executive system that handles the coordination and task execution issues