Abstract
In this paper, we present a method of easy communication with a robot similar to that which we use to communicate with each other when showing the way to some place. In traditional approaches to spatial learning, a mobile robot in unknown environments, tries to build metrically accurate maps in an absolute coordinate system, and therefore has to cope with device errors. Such learning is very important, however people view the structure of the environment topologically, the robot didn't. This view allows us to navigate without accurate measuring. When we direct someone to somewhere, we sometimes use a freehand map. If we could teach the robot to view things topologically, it could develop a map without physically visiting the environment.
We use a nonsymbolic method, which is a kind of neural network with self-organizing and plasticity properties, without metrical information, our nonsymbolic method build a simpler method of communication between people and robots, which approach that which exists between people and people. Our method presents a reduction of labor in teaching and the neeslessness of special teaching or drawing techniques.
With a simulation in which the robot explores corridor environments, we show its successful results in the face of random device error.
