Let It Roll - Emerging Sensorimotor Coordination in a Spherical Robot

Abstract

Self-organization and the phenomenen of emergence play anessential role in living systems and form a challenge toartificial life systems. This is not only because systemsbecome more life like but also since self-organization mayhelp in reducing the design efforts in creating complexbehavior systems. The present paper exemplifies a generalapproach to the self-organization of behavior which hasbeen developed and tested in various examples in recentyears. We apply this approach to a spherical robot drivenby shifting internal masses. The complex physics of thisrobotic object is completely unknown to the controller.Nevertheless after a short time the robot developssystematic rolling movements covering large distances withhigh velocity. In a hilly landscape it is capable ofmanoeuvering out of the basins and in landscapes with afixed rotational geometry the robot more or less adatps itsmovements to this geometry -- the controller so to saydevelops a kind of feeling for its environment althoughthere are no sensors for measuring the positions or thevelocity of the robot. We argue that this behavior is aresult of the spontaneous symmetry breaking effects whichare responsible for the emergence of behavior in ourapproach.