Extract

There have been multiple recent lines of quantitative inquiry into the correlations between spatial parameters of virtual enclosures, and the corresponding emotional responses in occupants. The application of such empirical approaches for occupant-specific predictive affective modeling and design assistance is, however, a very nascent domain. This paper outlines a design assistance workflow for rapid user-specific data collection and predictive affective analysis of enclosures in early stage design. For demonstration, 100 enclosures randomly generated along 9 spatial parameters – length, width, ceiling height, sill height, lintel height, no of windows, window position, total window width and wall hue – were presented to 5 subjects through cursor controlled displays as well as immersive VR environments, and were subsequently rated on an EmojiGrid, based upon the Circumplex Model of Affect. For design assistance, the framework allowed the designer to import any subject-specific dataset, and then applied K-Nearest Neighbour (KNN) regression algorithms to evaluate the affective impact of a designed test enclosure in real time, based on its spatial parameters and subject-specific training data. This research thus aims to integrate ‘subjective’ patterns of affective response into a computational framework, and open up possibilities for the ‘emotional customization’ of spaces.

Keywords: Affective Customization; Enclosure Analysis; Machine Learning; Design Assistance.

PDF

Download