Adapting principles of developmental biology and agent-based modelling for automated urban residential layout design

What is it about?

This research is about how to get computers to design things by themselves. By design, we mean coming up with new solutions from scratch, rather than modify parameters of existing solutions. The latter is more about optimisation. Computers are not creative and they're rigid so to get them create new designs that haven't not been prescribed by humans poses a major challenge. Our solution is nature inspired. It grows new solutions step by step, rather than tries to generate a whole new solution in one go. The use of 'adaptive cells' is the foundation of this approach. They act just like biological cells, i.e. they contain the same 'genes' and same developmental potentials but will change their physical shape according to where they are in the structure.

Featured Image

Why is it important?

This approach utilises the emergent behaviour to create much from little. Complex structures can be generated from a simple set of rules (genes) contained in adaptive cells. The complexity is a direct result of the interaction between adaptive cells. When human design things, we need to specify every parameter so the number of decision variables grows as the complexity increases. It obviously limits the level of complexity we can achieve. It is hard to image design something as complicated as a human using that approach. The fact that our genetic information can be contained by a DVD suggests that nature uses a different approach. The developmental approach attempts to solve the complicity problem by learning how nature uses the generative process to come up with novel structures that fit the environment.

Perspectives