| 摘要 |
FIELD: electricity. ^ SUBSTANCE: said control system contains a computer, put in a robot body fitted with trained neural networks, computer vision system, consisting of a television camera installed on the head of the robot, and a video signal processing unit. Tactile sensors are located on the surface of the robot hands and are connected to inputs of a microcontroller. A beacon light is put onto the index finger of the robot hand. The control method involves training three neural networks on carrying out tasks, related to seising and holding different objects, formation of final display access of relative position of the robot and the object and selection of their corresponding values of angles of deflection of the body, manipulators and robot hands, as well as degree of bending of the fingers and degree of curvature of palms. The computer then assigns the task, for example "grab" an object. The first neural network sends signals to the microcontroller with values of angles of deflection of the body and manipulators of the robot to as to touch the object. The second neural network adjusts these angles. The third neural network provides the microcontroller with values of angles of deflection of the robot hands, degree of bending of the fingers and degree of curvature of the palms. ^ EFFECT: more functional capabilities of the robot. ^ 4 cl, 10 dwg |
