My initial approach to generate floor plans in Rhino + Grasshopper was to divide a given perimeter into containers, and pack them with satisfying combinations of rooms. Researching tiling and bin packing has helped greatly, along with methods in computational geometry. A reductive approach has been very helpful in the preliminary stages of space planning, but much work is yet to be done to refine the algorithm so that a working floor plan is created.
As I develop the algorithm, I am often forced back to the beginning of a data stream to see if something more elegant could accomplish the same task. Working alone has given me flexibility to aggressively modify the algorithm in ways that would have faced more resistance in a team setting. I would love collaboration because personal eurekas can be harder to come by, but I have enjoyed the absence of a sunk cost that a team project might exhibit. In the spirit of starting over again, I’d like to cover some of the fundamental aspects to the algorithm.
A floor planning algorithm could empower designers, but could cease aiding the process in all kinds of ways. CAD has computerized previously analog drafting tools, greatly speeding up the drafting process. Computers, however, are capable of more; specifically, repetitive tasks in computation and analysis. I do not envision that a floor planning algorithm would run autonomously, but that it could provide real-time feedback from designer inputs, complementing the designer’s intuition and creativity with its calculated intuition.
Whether it is chocolate covered strawberries, cafe con leche, or the wonderful assortment of ingredients in paella, the right food combinations can taste so good. Combinations can also provide incredible insight for design and decision making. As I learn more about math, nature, and paella, I gain new perspectives on how they may influence my approach toward design and the process.