ABSTRACT
The dynamics of a general n-link robotic manipulator is expressed in the Lagrange-Euler formation as
M q C q G( )q ( )q q ( )q + τ τ=d (1)
where q = [ ], ,, R∈n T n
1 2 , q ∈R n, and q ∈Rn
are the vectors of joint position, velocity, and acceleration respectively. M( )q ∈ ×Rn n represents the inertia matrix; C( )q q ∈ ×Rn n denotes the matrix
1 INTRODUCTION
In the past decade, considerable research development has been achieved in the use of fuzzy inference system for the tracking control of robot manipulators when they suffer from structured and unstructured uncertainties such as load variation, friction, and external disturbances etc. In the earlier methods [1,2] the fuzzy rules for designing the fuzzy controller are built according to the designer experience and can not be varied once they are determined. By combing the learning ability of neural networks, fuzzy neural networks can overcome the shortcomings of the conventional fuzzy inference systems. Many adaptive fuzzy control schemes based on fuzzy neural networks have been developed for the robot manipulators to achieve accurate trajectory tracking and good control performance [3,4].
