The Problem of Maze Passage by the Intelligent Agents
Alexander B. Filimonov, Н. Б. Филимонов, V. Ju. Tikhonov · Mekhatronika Avtomatizatsiya Upravlenie · 2016
The problems of the analysis and passage of the unknown medium by wandering automata over it is the subject of the study of the theoretical computer science. The maze is very popular geometric model of such medium. The mazes as the artificial testing medium for the experimental research of traffic control algorithms of the autonomous mobile robots are widely used in robotics. The paper is dedicated to the problems of maze passage by robotic intelligent agents (IA). The most principle aspects of the algorithmization of mobile robots conduct may be study by means of computer simulation of maze search. The conception of virtual mazes and virtual robots living in them is discussed in the article. It is given the short review of the results of the theoretical researches of the final automata conduct in chess mazes. They permit to understand the possibilities of mazes passage by the reactive agents. The problem of going around of mazes by the IA occupies the central place in the article. The IA in contrast to the reactive agents are capable to analyze the habitat, to design and optimize its own conduct. In connection with it the statement of the maze problems is changed cardinally. The architecture of the IA which are able to decide the wide class of the maze problems is analyzed. The question of the structure of the situating models of IA conduct occupies the central place in the article. The simplest situating model of the agents' conduct is the static one and it is constructed on the basis of the productions "situation-action". But for the planning ofIA conduct it is necessary the dynamic situating model complemented with the event mechanism of the situations control. In this case the cycle of control by the agent conduct has the following stages: the analysis of the situation - the generation of local purpose - the planning of actions - the fulfillment of plan. The formal scheme of the structure of the mathematical models of the situating control processes and the planning of the actions of the IA is proposed. The considerations about the co-ordination of the agents' actions are reduced. The method of the optimization of the routes of the movement in the maze based on the reduction of the initial optimization problems for the checked maze to more simple problem of search of the shortest route of the weigh graphs is suggested.