Computing a Qualitative Representation for the Local Space

Yeap and Jefferies [1] show how a description of each local space is computed and by realising how these local spaces are connected, a cognitive map of one’s environment will emerge. However, roboticists have shown that it is difficult to compute such a map with precise metric information. In this paper, we show some preliminary ideas as to how a qualitative map could be constructed out of local spaces. The trick is to devolve each representation to make it less precise and then slowly learn more about each space so that it later becomes unique.