Modular Learning Schemes for Visual Robot Control

摘要: This chapter explores modular learning in artificial neural networks for intelligent robotics. Mainly inspired from neurobiological aspects, the modularity concept can be used to design networks. The main theme of this is explore organization, complexity and A robust architecture then developed position/orientation control a robot manipulator with visual feedback. Simulations prove that enhances capabilities learn approximate complex problems. proposed bidirectional avoids well-known limitations. Simulation results on 7 degrees freedom robot-vision system are reported show performances approach high-dimensional nonlinear problem. Modular thus an appropriate solution due limitations amount available training data, real-time constraint, real-world environment.