The article shortly introduces the Uncanny valley hypothesis and sums up some of the research performed in the field connected to it. Thereafter, it explores some possible new approaches in robot design. The main hypothesis is that pleasant human-robot interaction is based in the habituation of humans to this kind of interaction. This pleasant interaction can be accomplished by exploiting the human tendency to ascribe intentionality to even simple entities or by letting robots express emotion-like states through vocal communication. A possible risk of a new Uncanny valley phenomenon, from the view of artificial intelligence, is also described.