Evolvable case-based design: An artificial intelligence system for urban form generation with specific indicators

Abstract

This research proposes a design system that combines a case-based learning algorithm with a rule-based optimization algorithm to automatically generate and revise urban form prototypes based on historical cases and user requirements. The system aims to address the challenges of existing generative methods for urban forms, such as the lack of flexibility and organicity of rule-based methods and the insufficient manipulability and interpretability of the newest GAN-integrated case-based methods. It can help designers generate multiple solutions with specific indicators in the conceptual stage and has the potential to facilitate citizen participation in urban planning and design. This research demonstrates the feasibility and effectiveness of the system through a case study in Shenzhen. The research further extends the discussion about the application of the proposed system and the alternative evolution approach for the next generation of automatic design methods.