Date: October 14, 2026
Speaker: Claire Liang, Postdoctoral Fellow, MIT's Computer Science and Artificial Intelligence Laboratory
Title: Public space navigation without a map: From theory to the wild
Host: Kuan Fang
Abstract: Robots navigating public spaces need to adapt to dynamic and unpredictable environments. These public spaces are designed for intuitive human use, not robots. Robots in these settings cannot expect a priori information, such as maps, catered to its needs; Spaces, layouts, and conditions may change from moment to moment. A robot must act robustly even without knowledge of the full environment, and adapt when its prior beliefs about the environment are contradicted during deployment.
In this talk, I describe the development of a robot, from algorithms to physical system deployment, that leverages both innate qualities of environments designed for humans and bystander interactions to navigate intelligently without any prior knowledge of the space. I describe the full pipeline, including theoretical guarantees for navigation completeness to an in-the-wild study in a real public space, and conclude with my research vision to overcome hurdles to widespread and welcome adoption of robots in public.
Bio: Claire Liang is a CSAIL METEOR Postdoctoral Fellow in the Interactive Robotics Group (IRG) at the Massachusetts Institute of Technology (MIT), working with Professor Julie Shah. She got her Ph.D. in the Verifiable Robotics Research Group (VRRG) with Professor Hadas Kress-Gazit at Cornell University. Claire works in algorithmic human robot interaction with a focus on lightweight mathematical abstractions of human-intuitive spatial information. She approaches system development through collaborative participatory design and in-the-wild deployments to ensure robots that are not only successful at their tasks but are readily welcomed by the people around them.