Adaptive Neural Compensator for Robotic Systems Control

DANIEL GANDOLFO; FRANCISCO G. ROSSOMANDO; CARLOS M SORIA; RICARDO CARELLI

IEEE LATIN AMERICA TRANSACTIONS

IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC

Lugar: New York; Año: 2019

1548-0992

In the area of robotics systems, there are numerousapplications where robots are expected to move rapidly from oneplace to another, or follow desired trajectories while maintaininggood dynamic behavior. However, certain non-linearities,uncertainties in dynamics and external perturbations make thedesign of ideal controllers a complicated task in many situations.In this paper, we propose a control scheme that combines anominal feedback controller with a classical PD and a robustadaptive compensator based on artificial neural networks. Usingthis control scheme, it is possible to obtain a fully tunedcompensation parameters and a strong robustness with respect touncertain dynamics and different non-linearities, as well as to keepthe output tracking error bounded to values close to zero. In orderto show the performance of the proposed technique, a SCARA(Selective Compliant Articulated Robot Arm) type robot with twodegrees of freedom is considered in this case; but this controlproposal can be applied to different systems with dynamicvariations. The efficiency and performance of the control law isdemonstrated through simulation results and the stability analysisis carried out using Lyapunov´s theory.