We try to perform geometrization of psychology by representing mental states, “ideas,” by points of a metric space–mental space. Evolution of ideas is described by dynamical systems in metric mental space. We apply the mental space approach for modeling of flows of unconscious and conscious information in the human brain. In a series of models, Models 1-4, we consider cognitive systems with increasing complexity of psychological behavior determined by the structure of flows of ideas. Since our models are in fact models of the AI-type, one immediately recognizes that they can be used for creation of AI-systems, which we call psycho-robots, exhibiting important elements of the human psyche. Creation of such psycho-robots may be useful in the improvement of domestic robots. At the moment domestic robots are merely simple working devices (e.g. vacuum cleaners or lawn mowers). However, in future one can expect demand for systems which can not only perform simple work tasks, but also have elements of human self-developing psyche. Such AI-psyche could play an important role both in relations between psycho-robots and their owners as well as between psycho-robots. Since the presence of a huge numbers of psycho-complexes is an essential characteristic of human psychology, it would be interesting to model them in the AI-framework.