Nature AI Lab
Latest publications, guest lectures, conferences, and collaborations from the lab.
We hosted Jimmy Lee (Digital Architecture Research Institute, Wuhan University) for a guest lecture on AI methods for urban environmental research — covering urban morphology analysis, environmental simulation, and data-driven design.
Land Use Policy published our machine-learning framework assessing how urban greenery shapes urban vitality across space and time — integrating seasonal remote sensing (NDVI, BGCI, LST) with social-media vitality indicators across working days and holidays.
Sustainable Cities and Society published our scalable framework combining crowdsourced delivery-rider videos (EgoCity) with a custom vision-language model (LLaVA-PI) to measure street utilization and inclusiveness — only 4.9% of streets achieve both.
Our team joined the 65th ACSP Annual Conference (Oct 23–25, Minneapolis). Led by Mingze Chen, with Xiamengwei Zhang, Yidong Yang, and Xiao Wang taking part in sessions and discussions on research application and interdisciplinary methods.
We welcomed Guanhong Li (National University of Singapore) for a guest lecture on diffusion models and LLMs in urban design, cultural modeling, and planning evaluation.
Urban Forestry & Urban Greening published our comparative study of green-space inequality in New York and London, integrating multiscale spatial modeling (OLS, GWR, MGWR) with census and land-use data to reveal how income, race, education, and the built environment shape green accessibility.
Urban Climate published our study of urban scene types and the heat-island effect in NYC, using interpretable machine learning (SHAP, ALE, counterfactuals) to show how building height, vegetation, and socio-economic disparities shape land surface temperature.
Habitat International published our study on urban amenity equity for people with disabilities in London, using multiscale GWR to reveal disparities — especially in access to high-quality green spaces and commercial areas — and to call for disability-inclusive urban planning.
A visit to the Meta Boston office exploring the intersection of AI and urban landscapes — with thanks to Marcie Hogan for the warm welcome and a thoughtful tour, and inspired by the fusion of art and technology throughout the space.
Urban Forestry & Urban Greening published our COVID-era study of public-space visitation in Las Vegas, combining machine learning, GPS, and review data to compare the nonlinear effects of physical environment and human perception — and the divergent resilience of parks and commercial areas.
Our London study introduces a framework to quantify and map how pedestrians perceive street greenery, using space syntax metrics and crowdsourced imagery for more informed urban planning and design.
Health & Place published our systematic review of tools for measuring senior behavior in urban nature, highlighting health benefits and behavioral patterns.
Energy & Buildings published our systematic review of advancements in supervised machine learning for outdoor thermal comfort — across scales, applications, and data types — to refine predictive methods for urban climatic challenges.
Journal of Environmental Management published our scoping review of the “blue” habitat of urban and suburban areas — examining aquatic ecosystems, biodiversity challenges, and interdisciplinary methodologies for sustainable conservation.
We presented “Defining Urban Vitality Using Image-based, Text-based, and GPS Data” at the 2024 ACSP Annual Conference in Seattle, under the guidance of Dr. Keunhyun Park.