手先負荷推定ニューラルネットワークを用いたマニピュレータの軌道追従制御

概要

論文の詳細を見る
This paper presents a tracking control of robot manipulator. Recurrent neural network is used to estimate hand loads of manipulator and is trained by genetic algorithms. In this study, the angular acceleration of the manipulator is needed to estimate hand loads. These can be estimated using the angle, the angular velocity, and the control input to the manipulator. Layered neural network receives estimated hand load values from the recurrent neural network as a part of inputs and directly outputs the control input to the manipulator. This layered neural network, which is disposed parallel to the conventional P-D controller in order to consider the stability of the manipulator, operates as a non-linear controller. Tracking control of the robot manipulator by this system is simulated by computer. Simulation results verify the validity of this system.
明治大学の論文