A modular learning architecture for orienting a robot in a visual servoing task

A robust modular neural architecture is developed for the position/orientation control of a robot manipulator with visual feedback. Modular learning enhances the neural networks capabilities to learn and approximate complex problems. The proposed bidirectional modular learning architecture avoids the neural networks wellknown limitations. Simulation results on a 4 degrees of freedom robot are reported to show the performances of this modular approach to learn a high-dimensional nonlinear problem.