Data-Driven AI for Robotics (DAIR) at NVIDIA Research
NVIDIA's DAIR group investigates how robots can learn directly from human data including videos, motion capture, and demonstrations to acquire generalizable skills across tasks, embodiments, and environments. The group works at the intersection of generalist models and embodied AI.
Notable achievements
GEM (Generalist Model for Human Motion), Kimodo (Scaling Controllable Human Motion Generation)
Notable work
Recent publications
All papers →Matched by this lab's specialties (keyword overlap + direct affiliation)
Learning passive variable impedance skills for contact-rich tasks via conservative extended dynamical systems
Pingyun Nie, Jiexin Zhang, Tianxiang Jiang +4 more
Robotics and Computer-Integrated Manufacturing · 2027
DynaFLIP: Rethinking Robotics Perception via Tri-Modal-Dynamics Guided Representation
Jusuk Lee, Seungjae Lee, Jonghun Shin +6 more
2026
Robustness of Robotic Manipulation: Foundations and Frontiers
Yifei Dong, Zhanyi Sun, Lujie Yang +5 more
2026
Sequential Planning via Anchored Robotic Keypoints
Bryce Grant, Aryeh Rothenberg, Logan Senning +3 more
2026
Generative Learning as a Tool to Improve Perception of Emotional Body Motion Expressions
Huakun Liu, Miao Cheng, Xin Wei +6 more
2026
Learning flight navigation like a honey bee.
Matsiko A
Science robotics · 2026