Learning to Behave in Space: a Qualitative Spatial Representation for Robot Navigation with Reinforcement Learning