Papers

12

Total Citations

497

H-Index

7

About

Tengyu Liu is an emerging researcher at the forefront of robotic dexterous manipulation, 3D scene understanding, and embodied AI. His work addresses some of the most fundamental challenges in robotics: enabling machines to grasp, manipulate, and interact with objects with human-like dexterity and adaptability. Liu's most influential contribution is SceneDiffuser (165 citations), a unified diffusion-based generative model that integrates scene-conditioned generation, optimization, and planning in 3D environments — a significant departure from task-specific prior approaches. Equally impactful is his sustained focus on dexterous grasping, evidenced by DexGraspNet, a large-scale simulation dataset that has helped bridge a critical data gap in the field (85+ citations), and UniDexGrasp (94 citations), which achieves generalizable grasping across hundreds of object categories. His GenDexGrasp framework further advances cross-embodiment generalization across different robotic hand types. More recently, Liu has explored tactile sensing integration, multi-object grasping, and bimanual manipulation transfer, reflecting a broadening vision toward fully human-like robotic interaction. With over 490 cumulative citations and multiple high-impact publications in 2023–2025, Liu is rapidly establishing himself as a key voice in next-generation dexterous robotics and embodied intelligence research.

Research Focus

Key Achievements

7
H-Index
12
Papers
497
Total Citations
41
Avg Citations/Paper
🏆 Most Cited Paper
Diffusion-based Generation, Optimization, and Planning in 3D Scenes
165 citations · 2023
📈 Most Prolific Year: 2023 (6 Papers)
🤝 Key Collaborators: 47
🏛 Institutions: Beijing Academy of Artificial Intelligence, Beijing Institute for General Artificial Intelligence

Top Papers

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10

Key Collaborators

Contact & Links

Available for collaboration
Content generated · 14 days ago