Towards a computational spatial knowledge acquisition model in architectural space

Abstract. Existing research which is related to spatial knowledge acquisition often shows a limited scope because of the complexity in the cognition process. Research in spatial representation such as space syntax presumes that vision drives movement. This assumption is only true under certain conditions and makes these models valid only in specific scenarios. Research in human spatial cognition field suggests that the spatial information perceived by the individual is not equal to the visual appearance of the space, a straightforward way to rep-resent this cognition process quantitatively is lacking. Research in wayfinding usually assumes a certain degree of familiarity of the environment for the indi-vidual, which ignores the fact that the individual sequentially perceives infor-mation during wayfinding and the familiarity of the environment changes dur-ing the wayfinding process. In this paper, a conceptual spatial knowledge acquisition model for architec-tural space is presented based on the continuous spatial cognition framework. Three types of local architectural cues are concluded to relate common architec-tural elements to the continuous spatial cognition framework. With all relations in the proposed conceptual model quantitatively described, a computational model can be developed to avoid the aforementioned limitations in spatial rep-resentation models, human spatial cognition models and wayfinding models. In this way, our computational model can assist architects evaluate whether their designed space can be well perceived and understood by the users. It can help enhance the way-finding efficiency and boost the operational efficiency of many public buildings.