Michigan hosts 7th annual Learning for Dynamics & Control Conference

Approximately 300 researchers from around the world gathered in Ann Arbor on June 4–6, 2025, to attend the 7th annual Learning for Dynamics & Control (L4DC) Conference. The L4DC conference brings together engineers working on control theory, machine learning, and optimization, with the goal of updating the foundation of these fields to accommodate the increasingly massive sensing data sets being delivered in real time.

“We were very happy to host L4DC in Ann Arbor. L4DC is building a new interdisciplinary community at the intersection of controls, dynamical systems, machine learning, reinforcement learning, and optimization,” said Necmiye Ozay, the Chen-Luan Family Faculty Development Professor of Electrical and Computer Engineering (ECE) and general chair of the conference.

“What makes L4DC special is its perfect size and format for in-depth idea exchange and the rich cross-pollination between these fields. I personally enjoyed the technical program a lot and left the conference with many new ideas for future research.”

The proceedings included 16 oral presentations and 104 poster presentations of conference papers, from presenters across 23 countries. Keynote addresses were given by Yuejie Chi (Carnegie Mellon University), Nan Jiang (University of Illinois Urbana-Champaign), Asuman Ozdaglar (Massachusetts Institute of Technology), Marco Pavone (Stanford University/NVIDIA), Eduardo Sontag (Northeastern University), and Ambuj Tewari (University of Michigan Department of Statistics).

“The conference provided the perfect environment for conversations about people’s work, as well as their visions for the future of this field,” said Laura Balzano, associate professor of ECE and program co-chair of the conference.

“Given the interdisciplinary nature of L4DC, with people from machine learning, optimization, and control, there were many exciting new ideas being shared.”

The conference also included an opening reception at the Ford Robotics Building with six of the research groups providing tours and demonstrations. There were four additional tutorial sessions hosted, covering “Higher-Order Learning Dynamics in Games,” “Neural Networks in the Loop: a Tutorial on Robust Machine Learning for Control,” “The Scenario Approach: Data Science for Decision and Control,” and “Capabilities of Large Language Models (LLMs) in Control Engineering.”

“For robotics especially, bringing together diverse perspectives from control theory, machine learning, and optimization can accelerate breakthrough discoveries,” said Dimitra Panagou, associate professor of Robotics and L4DC program co-chair. 

“Hosting this gathering at Michigan allowed us to showcase the collaborative spirit that drives innovation in autonomous systems and robotics. The energy and creativity of the participants, combined with the high-quality research presentations, reinforced why this conference has become such an important venue for shaping the future of intelligent control systems.”

Ozay is an associate professor of both ECE and Robotics. In addition to her core appointment in ECE, Balzano is an associate professor of Statistics (by courtesy). In addition to her core appointment in Robotics, Panagou is an associate professor of Aerospace Engineering. The 2025 L4DC planning committee also included professors in the U-M Departments of Industrial and Operations Engineering; Aerospace Engineering; and Robotics, as well as the University of Oxford, Northeastern University, ETH Zurich, and Harvard University.