パラレルリンク形ダイレクトドライブロボットの学習制御

抄録

This paper describes the learning control method of a parallellink direct-drive robot manipulator. The acceleration error is used to correct the motion of the direct-drive motors. A theoretical method, which estimates the convergent limit of the learning control process based on the limit of a geometric series, is proposed. The validity of this proposed theory is proven through computer simulation ; it is then applied to the practical system of a parallel-link direct-drive robot manipulator. The convergent limit of a learning control system can be quantitatively estimated by the divergent coefficient which is defined in this study.