AI and Optimization: A Powerful Partnership

The National Science Foundation AI Institute for Advances in Optimization (AI4OPT) hosted Cathy Wu, Class of 1954 Career Development Associate Professor at the Massachusetts Institute of Technology, for a seminar titled “Towards AI-Assisted Optimization for Transportation System Design.” The talk highlighted how artificial intelligence can enhance optimization methods to create more efficient, sustainable, and reliable transportation networks.

Wu’s presentation centered on bridging the gap between artificial intelligence and traditional optimization, two fields that have historically developed in parallel. She described how combining the two can unlock new capabilities in system design, particularly for large-scale, complex networks such as city transportation systems.

Eco-Driving and Network-Wide Efficiency

A key focus of Wu’s talk was an eco-driving case study that applied deep reinforcement learning (RL) to optimize vehicle speeds across an entire traffic network. The study demonstrated that adjusting speeds at intersections could reduce emissions by 11% to 22% without compromising safety or traffic flow.

Overcoming the Limits of Deep Reinforcement Learning

Wu also underscored the limitations of pure deep RL in real-world applications. To address these challenges, her team is pioneering two complementary approaches: learning-guided optimization and contextual and transfer RL.

Building a Sustainable Future Through AI-Augmented Optimization

Together, these advances illustrate a path forward for AI-assisted optimization in transportation and beyond, offering tools to help planners and engineers design systems that are both adaptable and sustainable.

Wu, who holds a Ph.D. in electrical engineering and computer sciences from the University of California, Berkeley, leads interdisciplinary research at the intersection of AI, optimization, and engineering. Her work has earned numerous honors, including the NSF CAREER Award and the IEEE Intelligent Transportation Systems Best Dissertation Award.