Papers
7
Total Citations
155
H-Index
5
About
Weikang Wan is an emerging robotics researcher whose work sits at the intersection of dexterous manipulation, imitation learning, and generalizable robot policy learning. He is best known for **UniDexGrasp** (2023), a landmark contribution that achieved universal robotic dexterous grasping across hundreds of object categories — including unseen ones — by combining diverse grasp proposal generation with goal-conditioned reinforcement learning. The paper has accumulated 94 citations, establishing Wan as a notable voice in robot dexterity research. His broader portfolio reflects a consistent drive to make robots more adaptable: his work on generative adversarial self-imitation learning (2022) tackled category-level manipulation generalization, while **LOTUS** (2024) introduced continual imitation learning with unsupervised skill discovery, enabling robots to lifelong-learn new tasks efficiently. More recently, Wan has expanded into bimanual dexterous manipulation through **DexMimicGen** (2025), addressing critical data bottlenecks via automated demonstration generation, and contributed to **RoboVerse**, a unified platform for scalable robot learning. His exploration of spatial intelligence for vision-and-language navigation further signals an ambitious research agenda. Across his career, Wan's cumulative impact — over 150 citations — underscores his growing influence in building robots capable of truly generalist, real-world manipulation.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4
- 5
- 6
- 7