Intelligent system architectures - comparison by translation

System architectures are a central element to enable construction of increasingly complex software systems for intelligent artifacts. Scientific discourse about such architectures requires means to compare approaches and identify common directions. In this contribution we put forth the hypothesis that a meaningful comparison is possible only by translation to a common language. We present SYSTEMATICA 2D as such a common language, suitable for translation and comparison of architectures because it is both formal in its description and flexible in the range of systems which can be expressed. Main result of this contribution is the translation and comparison of three recent popular system architectures: the CogX George system[1], the ALIS3 system running on the Honda humanoid robot[2] and the iCub Cognitive Architecture[3]. Common patterns in all three systems become apparent from the two-dimensional ordering inherent in their SYSTEMATICA 2D translations and allow more founded discussions about architecture approaches and the exchange of concepts between real systems.