Land, Journal Year: 2025, Volume and Issue: 14(4), P. 786 - 786

Published: April 6, 2025

Thermal comfort in urban commercial spaces significantly impacts both business performance and public well-being. Traditional evaluation methods relying on field surveys expert assessments are often time-consuming labor-intensive. This study proposes a novel vision–language model (VLM)-based agent system for thermal assessment spaces, simulating eight distinct heat-sensitive roles with varied demographic backgrounds through prompt engineering using ChatGPT-4o. Taking Harbin Central Street, China as case study, we first validated accuracy ASHRAE scale evaluations of 30% samples (167 images) by 50 experts, then conducted simulations followed spatial interpretability analyses. Key findings include (1) significant correlation between VLM (r = 0.815, p < 0.001), confirming method feasibility; (2) notable heterogeneity across agents, demonstrating the VLMs’ capacity to capture perceptual differences among social groups; (3) analysis revealing higher eastern regions compared western central areas despite inter-role variations, consistency agents; (4) shade vegetation being identified primary influencing factors that contribute agent’s decision making. research validates VLM-based agents’ effectiveness evaluation, showcasing their dual capability replicating traditional while capturing group differences. The proposed approach establishes paradigm efficient, comprehensive, multi-perspective environments.

Language: Английский