Planning the behavior of an integrated unmanned aerial vehicle based on the division of complex tasks into subtasks in the state space

This work addresses one of the urgent problems of artificial intelligence related to the development of a model for the representation and processing of knowledge by an autonomous integrated unmanned aerial vehicle (robot) in the process of automatic planning of targeted activities in a priori undescribed conditions of a problem environment. The proposed model is based on the application of various scenarios in the form of frame-microprograms of behavior, frame-relations and frame-actions, as well as vague semantic networks that provide the representation of knowledge without reference to a specific subject area. This, in turn, allows an integrated unmanned aerial vehicle equipped with a manipulator to adapt after landing to a priori unknown operating conditions and to solve complex behavior problems on this basis. Two methods of template partitioning of complex behavior problems in the state space into simpler subtasks are used, the solution of which is determined on the basis of typical elements of knowledge representation and definition of fuzzy embedded isomorphism and equality of one vague semantic network to another. Planning procedures have been developed that allow the integrated unmanned aerial vehicle to efficiently transform the current situation of an a priori undescribed problem environment into a situation determined by the goal of behavior, and on this basis to organize goal-seeking activities in hard-to-reach and aggressive environments for humans.