神経回路により出力誤差のフィードバックを行う逆運動学問題の解法

概要

論文の詳細を見る
Many studies on the learning control of the robot arm have been conducted by using neural networks, The method that uses an acquired inverse-kinematics model of the arm by learning are popular. However, acquisition of the inverse-kinematics model has a number of drawbacks. Furthermore, a limited scale neural networks system has only limited precision. Errors still remains in the output of the inverse-kinematics model using the neural networks system. In this paper, a new method for solving inverse-kinematics problem using the learned inverse model of the linearized model as output feedback system is proposed. Two possible configurations of the system are presented. The use of linear adaptive systems including Kalman filter is also proposed for higher accuracy. The performances of the proposed methods are shown by numerical simulations.
日本ロボット学会,工業技術院機械技術研究所,東京大学工学部の論文
1995-01-15