Expert System-based Autonomous Mission Control for Unmanned Aerial Vehicle

UAV applications have witnessed a great leap during the last decade including aerial photography, surveillance, inspection, mapping and many other applications. Using UAVs has many advantages over manned aerial vehicles. Reducing costs and avoiding putting human lives in danger are two major benefits. Currently, most of the UAVs are remotely controlled by human operators, either by having Line of Sight between the operator and the UAV or by controlling it from a ground control station. This may be fine in short missions. However, manually executing long and boring missions adds much inconvenience on the human operators and consumes more human resources. In addition, there is always the risk of losing the connection between the UAV and the human operators which leads to unpredicted, and probably catastrophic, consequences. The objective of this work is to reduce this inconvenience by moving the decision making responsibility from the human operators to the mission control system mounted on the UAV. In other words, the target is to design an on-board autonomous mission control system that has the capability of making decisions on-board and in real-time. Expert system technology, which is a type of artificial intelligence, is used to reach the autonomy of the target UAV. Expert system has the advantage of dealing with uncertainty during the mission execution. It also makes the system easily adaptable to execute any mission that can be described in form of rules. In this thesis, the design, implementation and testing of the expert system-based autonomous mission controller (ESBAMC) is covered. The target mission used to prove the feasibility of the proposed approach is the inspection of power poles. Power pole insulator is autonomously inspected by capturing three pictures from three different points of view. The proposed system has been successfully tested in simulation. Results show the performance and efficiency of the system to make decisions in real-time in any possible situation that may occur during the execution of the considered mission. In the near future, it is planned to test the proposed system in reality