Newest faculty member teaches robots how to see, hear, and think

For the Department of Mechanical Engineering’s newest Assistant Professor, Frederike Dümbgen, engineering is about invention and understanding how the world works at a fundamental level. She blends math, physics, computer science and programming to connect computation with the physical world. 

Originally from Germany and raised in Switzerland, Dümbgen completed her undergraduate and graduate studies in mechanical engineering at EPFL. During that time, she became particularly interested in control theory and robotics, but because EPFL did not yet offer a dedicated master’s program in robotics, she designed her own path and focused her graduate studies on automatic control and robotics. She went on to complete a PhD in computer science at EPFL, during which her research focused on the way alternative signals, including sound, Wi-Fi, and radio signals, can be used for robotic perception and navigation. Inspired by bats and whales, her work emphasized sensor redundancy, much like how humans rely on hearing or touch when their vision is limited. 

Following her PhD she became a Postdoctoral Researcher at the University of Toronto Robotics Institute. There, she dove into optimization for robotics by creating algorithms that guarantee the quality of  solutions. From there, she spent one and a half years with Inria Paris in a similar role.

“Conventional solvers don’t come with guarantees,” she explained. “The more complex the problem, the harder it is to know if you’ve truly come to the right answer or if you should try again.”

At CMU, Dümbgen’s lab sits at the intersection of robotics and optimization. Her lab is designing algorithms that make existing robots, be it manipulators, legged robots, or autonomous vehicles,smarter, more capable, and adaptable. 

The exciting thing about robotics is that it can serve as a catalyst to put these advances to use for society. But to do this safely and reliably, a lot of fundamental research is still required.

Frederike Dümbgen, Assistant Professor, Mechanical Engineering

“The fields of optimization and AI are moving extremely quickly, and we constantly need to redefine what problems are solvable,” she said. “The exciting thing about robotics is that it can serve as a catalyst to put these advances to use for society. But to do this safely and reliably, a lot of fundamental research is still required.”

This semester, Dümbgen is teaching the graduate course Advanced Optimization for Engineering. In it, students start by building solvers for AI systems before moving on to constrained optimization with application to robotic kinematics and finally exploring global optimization. She is looking forward to incorporating a soft robot manipulator into the course as a safe, hands-on way for students to see their solvers solve real-world problems. 

Whether mentoring her students, designing new algorithms or exploring the next frontier in robotic perception, Dümbgen is driven by the thrill of uncovering how machines can truly understand and interact with the world around us.