Intelligent technologies for manipulation tasks

Abstract

The paper deals with technologies for the intelligent manipulators capable of making creative decisions automatically while operating. As examples of tasks classic puzzle problem and the table games like “Gomoku" or "Go" are considered. PlannerM (universal planning program) was developed and applied to solve the first task. The program was implemented on anthropomorphic kinematics model of a dynamic manipulator and on Visual Prolog and was realized on NAO robot (android robot). The second task was investigated from the point of view of the manipulator testing on the ManGo robot, its dynamic model, vision system and control system were tested. The logical moves choice program in the game “Gomoku" is designed in Prolog and C languages. Besides classical algorithms utilization, considerable attention has been given to the application of neural network paradigms for manipulators control. The elaborated robots passed experimental working off which allowed to draw conclusions about the software and hardware effectiveness in the implementation of various manipulators control algorithms.