Autonomous mission management for unmanned aerial vehicles

This paper presents an on-board architecture designed to enable an unmanned aerial vehicle to carry out an observation mission autonomously. The management performed by the architecture relates to the objectives of the mission. The response time to an external event does not need to be strictly defined; the architecture behaves like a soft real-time system. The mission of the vehicle is to observe several local areas in dynamic, partially known and unsafe environments. The vehicle must carry out an operation on each area while satisfying constraints related on the vehicle, the environment and the mission. The architecture contains a planning component, which computes the optimal sequence for the observation of the mission areas. The situation becomes degraded when the execution of the initial plan is no longer possible. The architecture makes it possible to react to these degraded situations; the calculation of a new plan takes into account the new constraints. The core of the architecture is an execution controller used in this application to specify the vehicle behaviour in the nominal and degraded situations and to manage the course of the mission. This tool is based on a Petri net formalism. The feasibility and the effectiveness of the architecture are checked by simulation with a simplified model of an aerial vehicle.