Goal-Directed Movement Generation with a Transient-Based Recurrent Neural Network Controller

Author

Reinhart, R. Felix ; Steil, Jochen J.

Author_Institution

Res. Inst. for Cognition & Robot. - CoR-Lab., Bielefeld Univ., Bielefeld, Germany

fYear

2009

fDate

22-26 July 2009

Firstpage

112

Lastpage

117

Abstract

We introduce a control framework based on a recurrent neural network for goal-directed movement generation. We exploit the network dynamics to implement a nonlinear task space controller. Efficient online learning and execution of the network makes the proposed approach adaptive and real-time capable. We achieve reliable and excellent generalization for the 7-DOF redundant PA-10 robot arm in simulation.

Keywords

learning (artificial intelligence); manipulator dynamics; motion control; neurocontrollers; nonlinear control systems; recurrent neural nets; 7-DOF redundant PA-10 robot arm; goal-directed movement generation; motion generation; network dynamics; nonlinear task space controller; online learning; transient-based recurrent neural network controller; Cognition; Cognitive robotics; Computer networks; Intelligent robots; Inverse problems; Machine learning; Neurons; Orbital robotics; Recurrent neural networks; Reservoirs; goal-directed movement generation; inverse kinematics; neural controller; recurrent neural network; reservoir computing;

fLanguage

English

Publisher

ieee

Conference_Titel

Advanced Technologies for Enhanced Quality of Life, 2009. AT-EQUAL '09.

Conference_Location

Iasi

Print_ISBN

978-0-7695-3753-5

Type

conf

DOI

10.1109/AT-EQUAL.2009.31

Filename

5231031