Cathy Wu
Biography
I’m an assistant professor at MIT LIDS, CEE, & IDSS. The main research goal of my group is to advance learning-enabled control and optimization. My group engages in use-inspired basic research; rather than apply known methods, we develop new methods to apply, necessitated by important societal problems with practical implications. The design of emerging mobility and logistics systems is exceedingly complex due to performance and reliability requirements, often inducing large families of difficult control and optimization problems. We call this the curse of variety. I focus on improving our basic understanding of the role that machine learning can play in overcoming this complexity. Our approach is to model emerging problems that suffer from the curse of variety, establish a baseline of understanding through both learning and non-learning approaches, and advance learning-enabled methods to overcome it. Our work has implications for vision zero, climate, and equity goals. More broadly, I am passionate about enabling policy-relevant research by pushing the boundaries of learning, control, and optimization.
I previously completed a postdoc with the Microsoft Research Reinforcement Learning group and my PhD in EECS at UC Berkeley. I received a BS and MEng in EECS at MIT. I have also spent time at OpenAI, Waymo, Dropbox, Facebook, and several startups.
Current project highlights: (a) One large effort we have is Project Greenwave, which leverages deep reinforcement learning to inform transportation decarbonization by mitigating carbon intensity of urban driving. (b) We are developing learning-enabled techniques for multi-agent coordination, motivated by applications in automated warehouses and connected and automated vehicles (CAVs). (c) We are exploring learning-enabled methods for combinatorial optimization, motivated by ever-evolving mobility systems. (d) We are developing statistical frameworks to address “how safe is safe enough?” to deploy autonomous vehicles (AVs).
Research funding: We are grateful for active support from Amazon, National Science Foundation (NSF), Mathworks, MIT Mobility Initiative (MMI), MIT Energy Initiative (MITEI), Microsoft Research, Utah Department of Transportation (UDOT), and MIT Research Support Committee.