Target Retrieval in Known Environment

Abstract

Target retrieval in known environment, when there is no straight visibility between robot and target point, is executing using various search systems, but each of them has their own lacks. If the target point forms surface, which gradually sinks from it, independently from form and amount of the obstacles, the gradient methods seems to be the best for the search. It is enough to know the direction of the gradient, if we want that the robot move directly to the target point in the shortest path. Gradient direction variations are established in order to form vector marks. Weight coefficients are assigned to vector marks. This coefficient is directly proportional to distance to target point. Using formed assumptions of this method, software tool, based on colored Petri nets, is created, which clearly demonstrated efficiency of the method. Independently from forms of obstacles, the shortest path to the target point is calculated. Ill. 6, bibl. 6 (in English, summaries in English, Russian and Lithuanian).