Bridging the gap between design space exploration and generative design interfaces: An exploratory study

This doctoral work aims to understand designers' search behaviour when navigating through a large number of design alternatives. The motivation for this research lies in previous studies on designers' search for design alternatives using different design media. With an increased computing capabilities and large screen displays, the opportunity to generate multiple designs has now become practical. However, due to the paucity of desired tool features, today's designers adapt ad-hoc techniques; such as opening two files side-by-side, layering designs for comparison, and saving versions manually. These techniques are rudimentary and have limited benefits and real costs when it comes to viewing multiple designs simultaneously and making sense of the overall design space. Recently some research has presented interfaces and system features for such exploration. However, before adopting any of these solutions, it is essential to understand the act of navigating and managing a large design space. The premise of our research is that, if designers can access and work directly with a large number of designs in an environment with new representations and tools as part of the design workflow, we expect new patterns and strategies to emerge and change the design process. What though are these new patterns and strategies? To answer this, I conducted a lab experiment with ten designers who were given one thousand design alternatives of an apartment building. The alternatives were produced using generative design techniques and were printed on index cards with an intention to discover how designers would engage with these designs in a controlled lab experimental setting. The results of the experiment revealed a cycle of design tasks performed repeatedly and patterns of spatial organizations to record designers' decisions. Based on the results from the experiment, I introduce high-level spatial metaphors and develop an interface prototype to support my analysis. I also present a comparison of existing CAD interfaces with respect to the proposed spatial metaphors.