A software architecture for Autonomous UAV Mission Management and Control

During the last decade, Uninhabited Aerial Vehicles (UAVs) have become increasingly popular in a large variety of application fields, both in the military and civil sectors. As a consequence, significant efforts have been spent on UAV research. The “race” towards increased UAV autonomy captured a considerable part of these efforts. At present, UAVs are capable of carrying out on their own a pre-planned mission. However, supervision is still needed, especially in highly dynamic environments where the flight plan might need to be changed. In this paper, a software architecture for Autonomous UAV Mission Management and Control is introduced. This is implemented using a combination of Soar intelligent agents and traditional control techniques, and is designed to be able to work in real-time. The system can generate a Flight Plan, execute it and update it during execution in order to deal with changes in the perceived environment. The architecture is outlined in detail, including the definition of necessary abstractions, prior to a thorough description of the system components. The simulation environment is described and simulation results are presented for both the Planner agent and the Execution agent. Conclusions focus on the suitability of this approach to autonomous UAV flight and on future developments planned.