Principle of Constructing Procedures for Planning Behavior of Autonomous Intelligent Robots Based on Polyvariable Conditionally Dependent Predicates

Abstract: We propose a principle of constructing procedures for planning the goal-directed behaviorof autonomous intelligent mobile robots of various purposes under underdetermined unstableoperating conditions. To build a knowledge representation model, its typical constructions havebeen developed in the form of implicative decision rules formed on the basis of polyvariableconditionally dependent predicates of various content. The structure of different-in-purposepredicates of this type, which can contain both polyvariables in the form of active fuzzy semanticnetworks and individual variables of the “object” and “event” types associated with certainconditions of the problem environment, is determined. It is shown that the use of active fuzzysemantic networks permits one to describe various situations and subsituations of the problemenvironment regardless of a specific subject area, as well as to determine in general terms therelationships that can be observed by an intelligent robot in the process of planning behaviorbetween its objects in the problem environment and the events occurring in it. Knowledgeprocessing tools have been developed at various stages of decision inference that allow constructingeffective behavior planning procedures providing autonomous intelligent mobile robots with theability to perform complex tasks formulated as a generalized description of the target situation ofthe problem environment. Upper bound estimates are found for the complexity of planningprocedures for purposeful behavior by an autonomous intelligent mobile robot built according tothe proposed principle of organizing tools for knowledge processing and decision inference. © 2022, Pleiades Publishing, Ltd.