Intelligent systems for autonomous aircraft

This paper introduces a mode-driven intelligent agent for the control of autonomous aircraft. The agent architecture uses pilot-type classifications of aircraft modes within all decision-making and reasoning processes. An essential component of the intelligent agent is a neuro-fuzzy inference engine that infers the current mode from sensor data and provides abstracted input for decision-making. The development of the intelligent system is structured around a nonlinear simulation of a conceptual V/STOL Uninhabited Aerial Vehicle. We are incrementally adding capabilities to achieve a multi-faceted surveillance and combat-capable intelligent pilot-substitute.