TLDR: Arizona State University is at the forefront of developing advanced AI tools designed to enhance real-world urban decision-making. Led by Assistant Professor Hua Wei, this initiative aims to overcome the challenges AI systems face in complex, unpredictable environments, focusing on applications like traffic control, public health, and energy management through adaptable and transparent AI.
Arizona State University (ASU) is spearheading innovative research into artificial intelligence, creating smarter tools to tackle the complexities of real-world urban environments. At the heart of this endeavor is Hua Wei, an assistant professor of computer science and engineering in the School of Computing and Augmented Intelligence, part of the Ira A. Fulton Schools of Engineering. His work focuses on bridging the critical ‘sim-to-real’ gap, where AI systems, despite excelling in simulations, often falter when confronted with the unpredictable and ‘messy’ dynamics of actual cities.
Wei’s project is dedicated to making AI tools more adaptable, transparent, and firmly grounded in reality. This involves addressing significant challenges such as noisy data, constantly shifting patterns, and the crucial need for human oversight in AI-driven decisions. ‘I want to help people make better decisions,’ Wei states. ‘Not just in city planning, but in daily life. Computer science can support that in a powerful way.’
His visionary approach has earned him a prestigious 2025 National Science Foundation Faculty Early Career Development Program (CAREER) Award. This recognition underscores the potential of his AI-powered project to revolutionize how cities manage essential services like traffic, public health, and energy systems, by providing intelligent, human-centered decision tools.
The research employs reinforcement learning, a sophisticated method where algorithms learn through trial and error to identify decisions that yield optimal long-term outcomes within urban systems. For instance, an AI-powered system can learn to dynamically adjust traffic light timings at intersections—controlling green, red, and yellow light durations—to minimize wait times and maintain smooth traffic flow, even in the face of sensor malfunctions or disruptive weather events. To rigorously test these methodologies, Wei and his team have developed a detailed test bed, a model of streets and traffic conditions based on the area around the Brickyard Building in Tempe, Arizona. They are also constructing a miniature replica of a city intersection, complete with sensors and traffic signals, for hands-on experimentation.
Also Read:
- AI’s Growing Influence on Personal Finance and Banking: An ASU Expert’s Perspective
- University of Florida Secures $4.7 Million Air Force Contract for Advanced AI in Military Planning
A key component of Wei’s research is uncertainty quantification, which enables AI systems to gauge their confidence in their own predictions. As a recipient of a 2024 Amazon Research Award, Wei is also advancing general-purpose algorithms for this crucial capability. By explicitly quantifying uncertainty, these tools empower human users to better interpret AI outputs and intervene when necessary, ensuring trustworthy and reliable decisions across high-stakes domains, from urban planning to healthcare.
