SME MS94-212

THIS PAPER PRESENTS A NOVEL APPROACH TO MANIPULATOR ARM CONTROL. THE SYSTEM CONSISTS OF TWO NEURAL NETWORKS. THE FIRST IS KINEMATICS NEURAL NETWORK AND THE SECOND IS DYNAMIC NEURAL NETWORK. UNLIKE CLASSICAL METHODS, THIS APPROACH DOES NOT REQUIRE A PRIORI KNOWLEDGE OF THE ROBOT DYNAMICS. A SELF-ORGANIZATION LEARNING SCHEME IS USED TO CONSTRUCT THE KINEMATICS NEURAL NETWORK AND A REWARD/PUNISHMENT LEARNING IS USED IN THE DYNAMICS NEURAL NETWORK TO MODIFY THE WEIGHTS OF THE NETWORK AND TO PRODUCE THE DESIRED TORQUE.

Product Details

Published:

06/01/1994

Number of Pages:

15

File Size:

1 file , 1.3 MB