A developmental constraint driven approach to developmental robotic hand-eye coordination

Robotic hand-eye coordination plays an important role in dealing with real time environment; and the learning procedure of this skill affects the fundamental framework of robotic cognition. Theories from developmental psychology have motivated the appearance of cognitive developmental robotics; however, several important features within infant developmental procedure have not been introduced into researches on developmental robotics. This paper proposes a new approach to robotic hand-eye coordination by imitating the developmental progress of human infant. The work employs a brain-like neural network system to learn hand-eye coordination. The learning procedure is driven by developmental constraints, the robot starts to learn under fully constrained conditions, when the system becomes saturated, a new constraint is assigned to the robot; the robot needs to learn this new conditions again, until the all contained conditions have been overcome. The work is supported by experimental evaluation, which shows that the new approach is able to drive the robot to own the hand-eye coordination ability autonomously, and the robot also behaves similar developmental progress with human infants´.