AI-driven computational kinematics
Represent motions, mechanisms, and constraints as learnable objects.
Research
From motion data to mechanical intelligence and real-world embodiment
Our research focuses on rigid body kinematics and machine learning for the design of robot mechanisms. We leverage analytical methods and machine learning based frameworks for the synthesis of N-bar planar mechanisms, and spatial linkages.
Core question
How can AI reason about the physical structures that make motion possible?
Three linked fronts
Human-centered robotic mechanisms
Design task-conditioned cobotic/assistive devices that match human motion.
Translation and education
MotionGen, SnappyXO, and physical AI learning.
This animation shows several mechanisms being designed by our AI-synthesis engine in MotionGen:
