Research
I equip robots with foundational motion skills that are broadly applicable to a wide range of tasks. The tools I develop are easily adaptable to new robots and environments, paving the way for future applications that demand significant mobility and control—whether it’s a construction robot, a cave explorer on distant planets, or a home robot that can help you move a couch!
|
|
Bridging the Sim-to-Real Gap for Athletic Loco-Manipulation
Nolan Fey, Gabriel B. Margolis, Martin Peticco, Pulkit Agrawal
Workshop on Robot Learning, ICLR 2025 (Oral)
Enhancing the sim-to-real transfer for a legged manipulator by calibrating actuator dynamics and refining whole-body control, enabling agile feats like throwing, lifting, and dragging
Paper
/
Website
/
X
|
|
A Learning-based Framework to Adapt Legged Robots On-the-fly to Unexpected Disturbances
Nolan Fey, He Li, Nicholas Adrian, Patrick Wensing, Michael Lemmon
International Conference on Learning for Dynamics & Control, 2024
A learning-based framework that allows a walking robot to stabilize itself under disturbances neglected by its base controller. We applied it to stabilize the MIT Mini Cheetah as it carried a box of water on its back.
Paper
/
Video
|
|
3D Hopping in Discontinuous Terrain Using Impulse Planning with Mixed-Integer Strategies
Nolan Fey, Robert Frei, Patrick Wensing
IEEE Robotics and Automation Letters, 2024.
By approximating each of the robot's stance phases to be impulsive, we enable a mixed-integer program to quickly plan consecutive hops for a quadruped between surfaces while avoiding obstacles.
Paper
/
Video (IEEE)
/
Video (YouTube)
/
Code
|
|