Every architectural design process starts with the schematic design phase, wherein architects have to satisfy a collection of adjacency constraints among spaces and dimensional constraints over each space element. Here, architects face a complicated problem. Some constraints contradict others; priorities may not be clear and the adjacency constraints grow exponentially as the number of rooms in a design problem increases. In large design problems, optimizing such a problem is a time consuming trial- and-error task that could benefit from computational assistance.
Among the different computational methods that have been used in optimization problems, artificial intelligence methods have shown a potential to produce novel optimized solutions. In this thesis, genetic algorithm, one of the powerful search methods in artificial intelligence, is used to create an intelligent prototype to be used in early phases of design. This prototype is able to generate alternative schematic designs to help the architects choose a direction for their design, while having a broad perspective about other good possibilities.