Michigan AI Symposium showcases cutting-edge advances in embodied AI

The event brought together over 300 AI experts for a day of networking and discussions around this evolving field.
Rada Mihalcea speaks at a black lectern on a wood-paneled stage. On the stage next to her are two robots, one low to the ground on two wheels, and one standing human-like on two legs.
Rada Mihalcea delivers opening remarks at the Michigan AI Symposium alongside two robot friends. Photo source: Levi Hutmacher, Michigan Engineering.

On Tuesday, October 15, over 325 researchers and practitioners convened at the University of Michigan for the 7th annual Michigan AI Symposium. This year’s theme, “Embodied AI,” drew thought leaders from around the country, who came together to share insights on the intersection of AI with physical and human interactions.

The event, organized by the Michigan AI Laboratory and co-chaired by professors Stella Yu and  Andrew Owens, featured presentations by leading experts, around 100 poster presentations, and several demos showcasing innovative AI applications. The symposium received generous support from sponsors including LG AI Research, DeepMind, Toyota Research Institute, Microsoft, Ford, P&G, Forvia Hella, Gladstone AI, Jane Street, eLab Ventures, and Unitree.

“We were delighted to host such a diverse and accomplished group of AI experts and users from both academia and industry,” said Yu. “The symposium not only highlighted groundbreaking research being done in this space, but also fostered valuable connections within the AI community.”

Provost McCauley shakes hands with a humanoid robot on Stamps auditorium stage
U-M Provost McCauley was welcomed onstage by two Unitree robots.

The symposium began with a welcome address by Rada Mihalcea, Janice M. Jenkins Collegiate Professor of Computer Science and Engineering and Director of the Michigan AI Lab. Her opening was followed by remarks from U-M Provost Laurie K. McCauley, who was greeted by two friendly Unitree robots, as well as an introduction by the symposium chairs.

The day’s talks offered compelling insights on topics ranging from value-aligned human-AI interaction to robotic sensorimotor learning, delivered by experts from institutions including Harvard, Stanford, MIT, and Carnegie Mellon.

The keynote speakers, in particular, provided a wealth of knowledge and inspiration. They included:

  • Karen Adolph, New York University, “How Behavior Develops” – Adolph shed light on the developmental processes that inform embodied interactions, highlighting how spontaneous, error-filled practice helps in building flexible behavioral systems.
  • Dieter Fox, Nvidia and the University of Washington, “Where is RobotGPT?” – Fox discussed the future of generalized AI for robotics and its potential implications, emphasizing the need for large-scale data to train robust AI models capable of real-world physical interactions.
  • Jitendra Malik, UC Berkeley, “When will we have intelligent home robots?” – Malik contemplated the timeline and future of AI in everyday life, covering advances in locomotion, navigation, and manipulation necessary for developing home robots.

Also providing enlightening information about exciting research in the field were invited talks by guest speakers Joyce Chai and Bernadette Bucher of U-M, Dinesh Jayaraman of the University of Pennsylvania, and Guanya Shi of Carnegie Mellon University.

You can view the full list of the day’s speakers and their abstracts on the symposium webpage.

A closeup (chest up) of Dieter Fox delivering his talk at a black lectern.
Keynote speaker Dieter Fox spoke about the the future of general AI in robotics.

The symposium also featured a poster and demo session, chaired by CSE PhD students Daniel Geng and Zilin Wang, where attendees exchanged ideas and research findings over lunch in Tishman Hall and the Stamps Auditorium lobby.

Outstanding poster presentations and demos were recognized with the following awards:

  • Best Technical Contribution
    Zixuan Huang, University of Illinois Urbana-Champaign, “Implicit Contact Diffuser”
  • Best Application
    Daniel Philipov, University of Illinois Urbana-Champaign, “Simulating User Agents for Embodied Conversational AI”
  • Best Embodied AI Poster
    Yixuan Wang, Columbia University, “CuriousBot”
  • Best Overall Poster
    Yinpei Dai, University of Michigan, “RACER”

“AI is progressing at a break-neck pace,” said Owens, “and this symposium reflected the wide range of research happening today, highlighting both foundational theories and practical implementations.”

This year’s symposium was made possible by local arrangement chair, the AI Lab’s Program Manager of Strategic Initiatives, and artistic chair, CSE Research Area Specialist Aurelia Bunescu. We would also like to thank the incredible staff, faculty, student volunteers, and sponsors who contributed to this year’s symposium.