ニューラルネットワークを用いた大型液晶基板搬送ロボットの学習制御(機械力学,計測,自動制御)

概要

論文の詳細を見る
Recently, a liquid crystal circuit board has become large accompanying with the increase in size of TV screen or display of personal computer. The robot system to carry the board must become also large. This study aims to develop an effective controller to drive such a robot which can realize both the high positioning accuracy and the response with little vibration. The vibration must be suppressed to avoid the damage to the board. The controller with constant gains like an usual PID controller may not be sufficient to realize the above requirements because of the parameter changes according to the position of the robot, the lack of structural stiffness and so on. To cope with these problems, a control strategy having an adaptability should be used. Also, for the convenience of practical use, it is desirable for the robot controller to have an auto-tuning ability. In order to construct a robot controller with the adaptability and the auto-tuning function, a learning controller using a neural-network is employed. In this paper, a hybrid type controller constructed by connecting an usual PID controller and a neural network in parallel is proposed. The PID controller can inhibit the unstable or oscillatory response at the beginning of the learning of neural network controller. The neural network controller can provide the robot system with the adaptability and auto-tuning ability. Some experiments confirm the effectiveness of the proposed controller as an practically available controller for the carriage robot for large liquid crystal circuit board.
2001-12-25

ニューラルネットワークを用いた大型液晶基板搬送ロボットの学習制御(機械力学,計測,自動制御)

スポンサーリンク

概要

著者

関連論文

スポンサーリンク