Advancing Safe, Reliable, and Collaborative AI for Tomorrow’s Autonomous Technologies
In the rapidly evolving fields of robotics, autonomous vehicles, and smart infrastructure, the question is no longer whether artificial intelligence can make decisions, it is whether those decisions can be made safely, reliably, and under uncertainty. A Fei Miao keynote typically centers on this precise challenge. Through a blend of control theory, machine learning, and rigorous mathematical modeling, Dr. Miao has emerged as a leading voice in the development of trustworthy, multi-agent AI systems for real-world applications.
Her keynote addresses do more than present technical findings. They frame a larger narrative about how embodied AI, systems that sense, learn, and act in physical environments, must be engineered with safety, resilience, and provable robustness at their core.
Fei Miao Keynote’s Facts
| Facts | Details |
| Full Name | Fei Miao |
| Current Position | Pratt & Whitney Associate Professor, School of Computing |
| Institution | University of Connecticut |
| Research Areas | Cyber-Physical Systems, Multi-Agent Reinforcement Learning, Robust Optimization, Control Theory |
| Undergraduate Education | B.S. in Automation (Minor in Finance), Shanghai Jiao Tong University (2010) |
| Doctoral Degree | Ph.D. in Electrical & Systems Engineering, University of Pennsylvania (2016) |
| Additional Degree | M.S. in Statistics, Wharton School, University of Pennsylvania |
| Dissertation Honor | Charles Hallac and Sarah Keil Wolf Award (Best Doctoral Dissertation, 2016) |
| Major Award | NSF CAREER Award (2021) |
| Keynote Highlight | IROS 2025 – “From Uncertainty to Action: Robust and Safe Multi-Agent Reinforcement Learning for Embodied AI” |
| Research Focus Areas | Autonomous transportation, Smart grids, Embodied AI, Resilient infrastructure |
| Editorial Roles | Associate Editor, IEEE Robotics and Automation Letters (RAL) |
| Conference Leadership | Area Chair, CPSWeek/ICCPS 2025 |
Early Academic Foundations: From Shanghai to Philadelphia
Fei Miao’s academic journey reflects both depth and distinction. She earned her Bachelor of Science in Automation, with a minor in Finance, from Shanghai Jiao Tong University in 2010. Her interdisciplinary foundation in engineering and finance would later inform her systems-level approach to optimization and decision-making.
She then moved to the United States to pursue graduate study at University of Pennsylvania, completing her Ph.D. in Electrical and Systems Engineering in 2016. While at Penn, she also earned a dual Master of Science degree in Statistics from the Wharton School, strengthening her expertise in uncertainty quantification and data-driven modeling.
Her doctoral work was recognized with the Charles Hallac and Sarah Keil Wolf Award for Best Doctoral Dissertation in Electrical and Systems Engineering, a distinction that foreshadowed her trajectory as a leading researcher in cyber-physical systems.
Academic Leadership at the University of Connecticut
Following postdoctoral research at Penn’s GRASP and PRECISE laboratories, Miao joined University of Connecticut in 2017 as a faculty member in the School of Computing. She now holds the Pratt & Whitney Associate Professorship and serves as a courtesy faculty member in Electrical and Computer Engineering.
At the University of Connecticut, Fei Miao leads a cyber-physical systems research group focused on autonomous transportation, smart grids, resilient infrastructure, and cooperative robotics, developing safer and more reliable embodied AI systems.
She is also affiliated with multiple interdisciplinary research centers, including the Institute for Advanced Systems Engineering and the Eversource Energy Center, reinforcing her commitment to translational research that connects theory with real-world impact.
The Intellectual Core of a Fei Miao Keynote
A Fei Miao keynote consistently highlights a central theme: how to design AI systems that remain safe and efficient under uncertainty.
Modern autonomous systems operate in environments characterized by incomplete information, adversarial risks, and communication limitations. Traditional machine learning approaches often assume idealized data conditions.
Miao challenges this assumption by integrating distributionally robust optimization, multi-agent reinforcement learning, game-theoretic control, and uncertainty quantification to design safer, more resilient AI systems.
Her work advances algorithms that account for sensing noise, communication delays, and unpredictable environmental changes, ensuring that safety guarantees hold even in worst-case scenarios.
From Uncertainty to Action: IROS 2025 Keynote
One of the most visible examples of a Fei Miao keynote was delivered at the 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) in Hangzhou, China. Her keynote, titled “From Uncertainty to Action: Robust and Safe Multi-Agent Reinforcement Learning for Embodied AI,” addressed a pressing issue in robotics: how multiple autonomous agents can cooperate safely while learning in dynamic environments.
In this address, she introduced frameworks for uncertainty-aware perception, provably robust multi-agent reinforcement learning, safe Shapley-value–based reward allocation, and distributed coordination among autonomous vehicles.
Rather than treating uncertainty as a secondary concern, her framework integrates it directly into learning and control design. The result is AI that does not merely adapt, it anticipates and safeguards against worst-case risks.
Collaborative Perception and CPS Safety
In 2023, Miao delivered another influential keynote at the “CoPerception: Collaborative Perception and Learning” workshop, co-located with ICRA. Her talk focused on safety, efficiency, and resiliency in cyber-physical systems.
A defining feature of her approach is the idea of collaborative perception, enabling connected vehicles and robotic agents to share information in ways that improve collective awareness. However, such sharing introduces new vulnerabilities, including unreliable communication or adversarial manipulation.
Miao’s research proposes uncertainty-quantified perception models, robust distributed learning strategies, scalable coordination mechanisms, and security-aware control architectures for networked systems.
Research Contributions in Mobility and Autonomous Systems
Beyond keynote platforms, Miao’s scholarship has reshaped several areas of intelligent infrastructure and autonomous mobility.
Distributionally Robust Urban Mobility
Her team developed data-driven models for ride-sharing and taxi dispatch systems that guarantee worst-case performance under demand uncertainty. By applying distributionally robust optimization, they created vehicle-balancing strategies that maintain service quality even when demand patterns fluctuate unpredictably.
These contributions, published in leading venues including ACM Transactions on Cyber-Physical Systems and Automatica, bridge theoretical rigor and operational impact.
Multi-Agent Reinforcement Learning with Safety Guarantees
Miao has also advanced robust deep multi-agent reinforcement learning algorithms. Unlike standard reinforcement learning approaches, her models:
- Explicitly account for sensing and communication uncertainty
- Incorporate formal safety constraints
- Provide performance guarantees under adversarial conditions
This work is particularly relevant for connected autonomous vehicles, where coordination errors can have physical consequences.
Awards and Professional Recognition
The themes explored in a Fei Miao keynote are grounded in a body of work widely recognized by the academic community.
In 2021, she received the National Science Foundation CAREER Award for her project on distributionally robust learning and control for connected vehicles. The NSF CAREER Award is one of the most prestigious honors for early-career faculty, recognizing both research excellence and educational impact.
Her papers have received multiple best-paper awards at top conferences, including ACM/IEEE ICCPS and AAAI workshops. She has also served in leadership and editorial roles, including as an associate editor for IEEE Robotics and Automation Letters (RAL).
Her nomination for the NSF Alan T. Waterman Award further underscores her standing among emerging leaders in science and engineering.
The Broader Impact: Trustworthy Embodied AI
What distinguishes a Fei Miao keynote is not only its technical sophistication but its philosophical clarity. She articulates a vision of AI systems that are:
- Provably robust rather than heuristically reliable
- Collaborative rather than isolated
- Safety-aware rather than performance-only optimized
- Grounded in control theory as well as data-driven learning
As autonomous systems expand into transportation networks, energy grids, and robotics applications, the stakes grow higher. Failures are no longer confined to digital errors — they translate into physical consequences.
Miao’s work addresses this reality head-on, emphasizing mathematical guarantees, worst-case reasoning, and resilient system design.
A Voice at the Intersection of Learning and Control
Across seminars at institutions such as Princeton, UC San Diego, Carnegie Mellon, and Berkeley, Miao continues to advance discourse around trustworthy AI. Her talks consistently integrate theoretical insights with applied case studies, making complex concepts accessible while preserving rigor.
The recurring message in any Fei Miao keynote is clear: AI must be engineered for uncertainty, not despite it. Safety and efficiency are not trade-offs but co-design objectives.
Conclusion: Redefining Safety in Multi-Agent Intelligence
In an era defined by rapid deployment of autonomous systems, the need for robust and secure learning frameworks has never been more urgent. A Fei Miao keynote offers more than academic insight, it provides a roadmap for designing embodied AI systems that operate safely in the real world.
Through her leadership at the University of Connecticut, her award-winning scholarship, and her global keynote platforms, Fei Miao continues to shape the future of multi-agent reinforcement learning and cyber-physical system design. Her work bridges theory and practice, ensuring that the next generation of intelligent systems is not only capable, but trustworthy.
As autonomous vehicles, smart cities, and distributed robotic networks become integral to modern infrastructure, the principles championed in a Fei Miao keynote, robustness, collaboration, and provable safety, are poised to define the next chapter of intelligent engineering.
