Navigation in evolving robots: Insights from vertebrates.

Mobile robots navigation is a broad topic, covering many dif- ferent technologies and applications. It is possible to draw inspiration for robot navigation from vertebrates. Reviewing literature on vertebrates, it seems clear that they navigate by elaborating substantially two kinds of spatial information: geometric (environmental shape, distance from landmarks) and non geometric (colors, smells). In this paper we try to understand how these cues can be used by small populations of mo- bile robots in environments reproducing the main features of experimen- tal settings used with vertebrates. The robots are controlled by neural networks, whose evolution determines robot navigation behaviour. We analyze how the arti¯cial systems use these information, separately or jointly, and how is it possible to obtain mobile robots that exploit e®ec- tively geometric and non-geometric information to navigate in speci¯c environments.