Research

Robots are leaving the lab. Before we let them, we should be able to say what they will and will not do. My group builds the math that makes that possible, and makes it fast enough to run on the robot.

The thread through our work is structure. Whether a robot is building a map from cameras and lidar, estimating its own state, modeling a cable or a human body, or deciding how to move, the problem underneath is usually nonlinear, high-dimensional, and uncertain. We look for the representation that makes it tractable, such as a lifted linear model, a Lie group, or a physics-based model, and then solve it fast and, where it matters, with a guarantee. Learned models plug into that structure rather than replacing it.

We test everything on hardware: humanoids, quadrupeds, drones, robot arms, and robots that handle cables and other deformable objects. See our research.

Biography

Ram Vasudevan is a Professor of Robotics and Mechanical Engineering at the University of Michigan, where he directs the ROAHM Lab and serves as Associate Chair of Graduate Studies in Robotics. He received a BS in Electrical Engineering and Computer Sciences and an Honors Degree in Physics in 2006, an MS in Electrical Engineering in 2009, and a PhD in Electrical Engineering in 2012, all from the University of California, Berkeley. He was a postdoctoral associate in the Locomotion Group at MIT from 2012 to 2014, and joined Michigan in 2015.

His honors include an NSF CAREER Award, an ONR Young Investigator Award, Michigan's 1938E Award, and the Michigan Robotics Departmental Faculty Award. Papers from his group have won best paper awards at IEEE ICRA, IEEE Robotics and Automation Letters, ASME DSCC, IEEE BioRob, and IEEE OCEANS, and have been best paper finalists at Robotics: Science and Systems.