Modelling the immune system: the case of situated cellular agents