Learning Turn And Travel Actions With An Uninterpreted Sensorimotor Apparatus

We demonstrate a learning method by which a mobile robot may analyze an initially uninterpreted sensorimotor apparatus and produce a useful characterization of its set of actions. By "initially uninterpreted" we mean that the robot is given no knowledge of the structure of its sensory system nor of the effects of its actions. It merely sees and produces vectors of real numbers. We apply the method to the case of a simulated robot with an array of 16 range finders and a motor apparatus with which it can make combinations of turning and advancing actions. The robot learns a set of primitive actions allowing it to make pure turns (both clockwise and counterclockwise) and pure travels. We believe our approach is robust and will apply in general to sensory systems which may be used for motion detection (such as arrays of photoreceptors or range finders)