Teaching New Tricks: How Scout the AI Dog is Advancing Research at the School of Statistics
University of Minnesota’s School of Statistics, the future of AI is taking four-legged strides into the real world. Led by Associate Professor Jie Ding and Ph.D. student Fangqiao Tian.
What inspired you to acquire the AI dog?
We were interested in having a real physical platform for testing research ideas beyond simulation. Many AI methods look promising in controlled digital environments, but the real challenge is whether they can remain reliable when facing uncertainty, changing conditions, and physical interaction with the world. The robot dog gives us a way to study these questions directly, while also creating new opportunities for hands-on student teaching and research training.
“What excites our students most about this project is that it lets us study AI in a real physical setting. A robot is no longer just solving a problem on a screen. It has to move, observe, adapt, and respond to whatever the world gives it.” Jie Ding
What specific research are you conducting, and what are you hoping to learn or teach with the dog?
One of our current interests is how AI systems can learn quickly from limited human demonstrations and then adapt in unfamiliar real-world environments. For example, a robot dog used for delivery or assistance may need to move across sidewalks, grass, ramps, stairs, or crowded hallways which are difficult to fully capture in lab data alone. Our research asks how such a system can detect environmental changes, recognize when it is uncertain, and adjust its behavior in a robust and reliable way.
Where do you see the future of AI and robotics heading?
We think AI is steadily moving from cyberspace into the physical world. In the future, robots will likely become much more interactive, practical, and integrated into everyday life. Just as importantly, we expect physical robots to become easier to operate: not merely as machines that execute commands, but as intelligent systems that can handle terrain, obstacles, and tasks they weren't explicitly trained on.
“What I find especially exciting is the opportunity to connect human guidance with real-world robot learning. The robot dog gives us a platform to study how AI can adapt quickly and safely when the environment is uncertain and constantly changing.” Fangqiao Tian
Does the dog have a name?
We currently call it Scout. The name reflects both exploration and learning, which fits the spirit of the project well.
One point that may also be worth highlighting is that this project helps connect foundational AI research with hands-on education. It gives students a concrete way to think about how learning, uncertainty, perception, and decision-making come together in a real system.
About the researchers
Jie Ding is an associate professor in the School of Statistics at the University of Minnesota. His research interest is in the foundations of statistics and artificial intelligence. His work has been recognized by the NSF CAREER Award, the ARO Young Investigator Award, and other honors.
Fangqiao Tian is a PhD student in the School of Statistics at the University of Minnesota. Her current research focuses on embodied AI, with an emphasis on how AI can learn from human guidance and interact effectively with the physical world.
We want to acknowledge that this work is supported in part by gift funding from Cisco.
