Papers
15
Total Citations
885
H-Index
8
About
Jan Kautz is a prominent computer vision and robotics researcher whose work spans 3D scene understanding, robot learning, and embodied AI. Best known for foundational contributions to perception and reconstruction, Kautz has shaped how machines interpret and interact with the physical world. Among his most influential contributions is **DexYCB** (2021, 250+ citations), a benchmark dataset for hand-grasping analysis that has become a standard reference for 6D object pose estimation and keypoint detection research. His **PlaneRCNN** (2019, 240+ citations) introduced a landmark deep learning architecture capable of detecting and reconstructing piecewise planar surfaces from a single RGB image — a significant leap for scene geometry understanding. **DeepGMR** (2020, 247+ citations) tackled the longstanding challenge of point cloud registration using latent Gaussian mixture models, offering robust solutions under noisy and large-transformation conditions. Kautz has also advanced sim-to-real transfer in robotics and domain adaptation through synthetic data, addressing real-world deployment challenges. His most recent work on humanoid whole-body control (**HOVER**, 2025) and legged robot navigation (**NaVILA**, 2025) reflects a continued push toward versatile, language-guided embodied agents. Across fields, his research consistently bridges rigorous perception methodology with practical robotic applications.
Research Focus
Key Achievements
Top Papers
- 1DexYCB: A Benchmark for Capturing Hand Grasping of Objects250 citations · 2021
- 2DeepGMR: Learning Latent Gaussian Mixture Models for Registration247 citations · 2020
- 3PlaneRCNN: 3D Plane Detection and Reconstruction From a Single Image240 citations · 2019
- 4
- 5
- 6DeepGMR: Learning Latent Gaussian Mixture Models for Registration16 citations · 2020
- 7NaVILA: Legged Robot Vision-Language-Action Model for Navigation13 citations · 2025
- 8HOVER: Versatile Neural Whole-Body Controller for Humanoid Robots12 citations · 2025
- 9DexYCB: A Benchmark for Capturing Hand Grasping of Objects7 citations · 2021
- 10