Papers
7
Total Citations
175
H-Index
6
About
Weihao Yuan is a robotics researcher whose work spans manipulation planning, tactile sensing, and 3D perception, with a particular focus on enabling robots to interact intelligently with objects in unstructured environments. His most recognized contributions lie in nonprehensile manipulation — the art of moving objects without grasping them — where his pioneering work on deep reinforcement learning combined with simulation-to-reality transfer (54 citations) demonstrated that robots could learn robust pushing behaviors entirely in simulation before deploying them in the real world. Building on this, Yuan extended these ideas through Monte Carlo Tree Search for multi-object planar sorting tasks (51 citations), offering a principled planning framework for arranging densely packed objects. His research also ventures into whole-arm manipulation for handling large, bulky objects (25 citations) and vision-based tactile sensing for detecting and predicting contact events during grasping (20 citations). More recently, Yuan has pushed into open-vocabulary object pose and size estimation (2024), applying natural language descriptions to novel object categories — a significant step toward generalizable robot perception. Spanning both learning-based and model-based approaches, Yuan's body of work reflects a broad and technically rigorous commitment to advancing real-world robotic dexterity.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4
- 5Open-Vocabulary Category-Level Object Pose and Size Estimation11 citations · 2024
- 6
- 7Active Perception with A Monocular Camera for Multiscopic Vision3 citations · 2020