ITS Berkeley

Accurate modeling of human mobility is critical for tackling urban planning and public health challenges. In undeveloped regions, the absence of comprehensive travel surveys necessitates reconstructing mobility networks from publicly available data. Here we develop neuroGravity, a physics-informed deep learning model that reliably reconstructs mobility flows from limited observations and transfers to unobserved cities. Using only urban facility and population distributions, we find that neuroGravity's regional representations strongly correlate with socioeconomic and livability status,...

Lithium-ion batteries are often retired while still retaining 70–80% of their rated capacity, creating economic and environmental challenges across their supply chain. Although reuse, recycling and remanufacturing offer alternatives to recover this otherwise under-utilized value, their implementation is hindered by the lack of reliable data on battery condition at retirement, making it difficult to determine whether, when and how these alternatives should be applied. In this Perspective, we discuss how artificial intelligence (AI) can help to overcome data barriers by enabling adaptive,...

Emerging evidence suggests that extreme heat elevates road-traffic injuries, undermining international road safety efforts like Vision Zero as global warming intensifies. However, the mechanisms underlying heat-related crashes remain poorly understood, with limited research linking heat exposure to specific crash types that may be driven by heat-induced unsafe road behaviors. Here, we analyzed temperature data and police-reported crash records—including detailed crash scene information—from 177 California cities (2012–2023) using a time-stratified case-crossover design, examining...

We introduce (U)NFV, a modular neural network architecture that generalizes classical finite volume (FV) methods for solving hyperbolic conservation laws. Hyperbolic partial differential equations (PDEs) are challenging to solve, particularly conservation laws whose physically relevant solutions contain shocks and discontinuities. FV methods are widely used for their mathematical properties: convergence to entropy solutions, flow conservation, or total variation diminishing, but often lack accuracy and flexibility in complex settings. Neural Finite Volume addresses these limitations by...