フィードバック制御, フィードフォワード制御の適用によるサッカー動作の生成手法 A Method Applied for Soccer's Behaviors Using Proper Feedback and Feedforward Control
As a practical way of athletic Intelligence, we choose the mobile robot which plays soccer to be a theme, and are advancing the research. Then, the dribbling and the shooting actions were achieved by using single robot Nomad200.<BR>The purpose of this research is in the relation of the control of feedforward (FF) and feedback (FB) to the achievement of these two kinds of operations respectively, and clarifying the problem how it is necessary to combine.<BR>We clarify the combination through operation (shoot and dribble) two kinds of experiments. For the example of kicking the rolling ball, we check whether the robot should have to follow the ball by feedback or move to the position by feedforward which is presumed by the change in the position of the ball.<BR>We took the standpoint where the designer has to tune the movement of the robot carefully. The robot first has the target to move so that the robot do actions properly and should repeat the re-calculation of the target.<BR>We explains how robot should set the target in the time series or how to approach the target in addition.<BR>Dribbling is achieved by feedback control so that the ball keeps existing on the straight line by which this target is connected with the robot. Next, the meandering dribbling is made to be done to the example of the operation that the target changes to the time series. For instance, the action that robot carries the ball to a target between the enemies. In this dribbling, robot first keeps maintaining the ball in front of the body. In a word, it is necessary to decide the direction which should move in the general situation adjusting the rolling condition of the ball.<BR>Because the each targets is different from a geostationary ball and the rolling ball, robot switches the shooting actions dependent on the conditions of the ball. To kick a geostationary ball to the direction, robot only has to set the target in the back of the ball seen from the direction. If the ball rolls, the target should also shifts. Then, the robot calculates and renews the target to follow the ball. If the robot follow the ball only by feedback control, robot cannot kick the rolling ball because the delay is taken in the movement. Robot presumes the position where the ball rolls beforehand, and sets the moving target. The target is presumed from the movement of the ball, and the forestalled technique is needed. Robot can move at a stroke earlier than ball moves to the target position for the feedforward control and wait for the arrival of the ball.
日本ロボット学会誌 16(6), 807-815, 1998-09-15
The Robotics Society of Japan