Planning Purposeful Activities Autonomous Intelligent Robot with Knowledge Update in Short-Term Memory

Abstract

The main problems associated with the creation of autonomous intelligent robots capable of performing various complex tasks in a priori undescribed unstable problematic environments, based on the processing of knowledge presented in an abstract way, are outlined. To store typical elements of an abstract knowledge representation model, the article recommends using long-term and short-term memory. Long-term memory with associative search and data retrieval is designed to permanently store information necessary for planning a variety of purposeful activities that provide the robot with the ability to solve various complex behavioral tasks. In short-term memory, submodels of knowledge representation are entered from long-term memory, which are necessary for solving the current task of a certain type in the short term, related to the fulfillment of the task formulated for the autonomous intelligent robot. At the same time, with each change in the type of the current task of behavior being solved by an autonomous intelligent robot, a corresponding update of knowledge stored in short-term memory is simultaneously carried out. Original constructions of typical elements of the model for representing abstract knowledge in the form of various behavioral skills, set regardless of a particular subject area, have been developed. This approach to building a knowledge representation model allows autonomous intelligent robots to adapt to the current operating conditions and, on this basis, organize purposeful activities in complex unstable problematic environments. Various tools and rules for processing abstract knowledge are proposed, which endow autonomous intelligent robots with the ability to eliminate the differences between the current and target situation of the problem environment both in terms of the values of structurally equivalent relations of the same name in them, and in the current states of objects in the environment. This, in turn, makes it possible to create intelligent problem solvers for autonomous intelligent robots for various purposes, capable of performing complex tasks in unstable a priori uncertain conditions of a problematic environment.