Self-development of motor abilities resulting from the growth of a neural network reinforced by pleasure and tensions

We present a novel method of machine learning toward emergent motor behaviors. The method is based on a growing neural network that initially produces senseless signals but later associates rewarding signals and quasi-rewarding signals with recent perceptions and motor activities and, based on these data, incorporates new cells and creates new connections. The rewarding signals are produced in a device that plays a role of a "pleasure center", whereas the quasi-rewarding signals (that represent pleasure expectation) are generated by the network itself. The network was tested using a simulated mobile robot equipped with a pair of motors, a set of touch sensors, and a camera. Despite a lack of innate wiring for a useful behavior, the robot learned without an external guidance how to avoid obstacles and approach an object of interest, which is fundamental for creatures and usually handcrafted in traditional robotic systems