Evolving lifelong learners for a visually guided arm

Abstract

The primary objective was to develop a fast learning dynamic controller for uncalibrated visual guidance of a robotic arm. A combination of neural networks learning with an evolutionary method allowed for the study of the interaction of the two techn