Visual predictive control architecture based on image moments for manipulator robots

Abstract

Increasing performances of nonlinear predictive control techniques for visual servoing systems, by using image moments as visual features, is the main goal of this paper. An local model based predictor is developed and the cost function is constructed using a image moments based reference trajectory. A new type of visual predictive control architecture is designed and a simulator is developed. Simulation results for a 6 d.o.f eye-in-hand scheme are revealed.