Agent‐Based Methods for Fabrication‐Oriented Design of Adaptive Modular Prefab Concrete Construction

This project develops computational design methods for the modularization of digitally fabricated concrete elements. By integrating fabrication constraints and structural feedback into an agent‐based model for early design stages it addresses a central challenge in industrialized construction: translating the formal freedom and structural capabilities enabled by digital fabrication processes with concrete into discrete, producible building modules. The presented method models each potential module as an autonomous agent that negotiates its size, shape, and position in a goal‐oriented manner within a volumetric representation of a given design geometry. Through this behavioral simulation process, feasible modularization variants that balance material efficiency, structural logic, and fabrication requirements are identified. Behavior weights are optimized through reinforcement‐learning based policy optimization, thus augmenting manual parameter tuning with data‐driven search. By integrating various design objectives within one consistent interactive simulation process, this project presents a design method to foster flexible and sustainable precast concrete construction from digital fabrication methods.