A STUDY ON THE QUANTIFICATION OF A FLOOR PLAN USING NPCS WITH NEEDS BASED AI IN A 3D SIMULATION

In recent years, AI has been increasingly utilized in the construction industry to address labour shortages and improve efficiency.While AI applications in construction management and site operations are advancing, their use in architectural design remains limited.This study proposes a novel method for evaluating floor plans quantitatively by simulating NPC (Non-Player Characters) within a 3D environment.Using a BIM-based CAD software and game engine, we developed a system that enables NPCs to navigate a 3D floor plan and interact with furniture and spaces.The system integrates need-based AIdriven NPC behaviours to assess spatial efficiency and usability.Data is transferred between CAD and the simulation environment via an external data-sharing platform, allowing real-time updates and automated synchronization.NPCs are designed with human-like needs and behaviour patterns, implemented using an AI model inspired by an existing need-based decision-making framework.Each NPC autonomously moves through the environment, interacting with designated objects based on predefined parameters such as hunger, work tasks, and movement efficiency.This allows for quantifying accessibility, privacy, and spatial bottlenecks.The study conducted simulations on simple residential and office layouts, measuring average travel distances and comparing movement complexity through a calculated movement distance ratio.Results suggest that this method provides valuable insights for optimizing spatial design, particularly in larger environments such as office spaces.Future work aims to refine the system by incorporating room-specific usage frequency analysis, subjective user-experience modelling, and diverse NPC profiles to simulate different lifestyles and work behaviours.This research contributes to advancing AI-based architectural evaluation tools, bridging the gap between traditional CAD-based design and real-world usability assessment.