Preliminary communication
Applied Soft Computing Robot Motion Control
Andreja Rojko
Riko Šafarič
Karel Jezernik
Abstract
This paper considers the problem of the robot motion control in the presence of the major uncertainties such as it is varying load. Efficiency of one conventional and two soft computing model based control algorithms is investigated and compared trough the results of application on a direct drive robot. First control algorithm is a conventional computed torque based on the Lagrangian dynamic equations. Second method is a computed torque alike control with an adaptive fuzzy logic system that replaces Lagrangian model, and third is a continuous sliding mode control with an artificial neural network instead of the dynamic model. Both soft computing methods give excellent results, while inefficiency of the computed torque control confirms the disadvantages of the conventional model based motion control approaches.
Keywords
adaptive fuzzy logic system; computed torque control; neural network; robot control; sliding mode
Hrčak ID:
18375
URI
Publication date:
15.11.2007.
Visits: 1.779 *