On-Line Identification of a Robot Manipulator Using Neural Network with an Adaptive Learning Rate
| Authors: | Velagic Jasmin, Faculty of Electrical Engineering, Bosnia and Herzegowina
Hebibovic Mujo, Faculty of Electrical Engineering, Bosnia and Herzegowina
Lacevic Bakir, Faculty of Electrical Engineering, Bosnia and Herzegowina |
|---|
| Topic: | 4.3 Robotics |
|---|
| Session: | Robot Control I |
|---|
| Keywords: | fuzzy controller, neural network, on-line identification, robot dynamics |
|---|
Abstract
This paper proposes an extention of neural network identification capabilities for on-line identification of a nonlinear closed-loop control system. The neural network is trained on-line using the backpropagation optimization algorithm with an adaptive learning rate. The optimization algorithm is performed at each sample time to compute the optimal control input. The results confirm the effectiveness of the proposed neural network based identification sheme and control architecture.
