This paper presents an implementation of functional modules for intermixed planning and execution of an observation mission for an UAV. The mission is a mean altitude navigation in a 3D partially known and dynamic environment. The mission plan includes operations such as area survey, outline following and object search, defined off-line by an operator. It must satisfy constraints due to the vehicle, the environment and the mission. The ProCoSA tool is used to specify the UAV desired behaviour through hierarchical Petri nets, to check the behaviour of the UAV during the execution and to communicate with computational sub-systems of the onboard system architecture. One highest level Petri net describes the observation mission : first off-line planning and take-off, then navigation to mission point and operation achievement if it exists, finally return-to-base and landing. Secondary generic Petri nets specify the different phases of the mission execution. System replanning capabilities are activated on-line if current trajectory or itinerary is no more consistent with prediction. The main external sub-system optimises the 2D itinerary and 3D associated trajectories given the mission and its constraints. A second sub-system computes the guidance sent to the modelled UAV. A third one centralises dynamic information. First simulation tests on a 200kmx 140km zone with all types of operations and a simplified model of a mean altitude UAV highlight the feasibility of the proposed structure and the reactive capabilities of intermixed planning and execution architecture in nominal and degraded situation.
展开▼