Abstract

This paper presents a comprehensive framework for machine learning-driven generative design tools that utilize reinforcement learning (RL) and evolutionary algorithms (EA) to optimize building layouts. This research approach explores thousands of design permutations while simultaneously optimizing for cost efficiency, material utilization, and structural integrity under various engineering constraints. The proposed system enables rapid prototyping and automated selection of optimal building configurations, reducing design time by 75% while improving structural performance by 23% compared to traditional methods. Experimental results demonstrate the effectiveness of the hybrid RL-EA approach across multiple building types and constraint scenarios.