Puhao Li

Beijing Institute for General Artificial Intelligence, Beijing Academy of Artificial Intelligence

Papers

Total Citations

H-Index

About

Puhao Li is an emerging researcher at the intersection of robotics, computer vision, and 3D scene understanding, with a focus on enabling intelligent, human-like manipulation in robotic systems. His most recognized contribution is **DexGraspNet**, a large-scale simulation-based dataset for robotic dexterous grasping that has rapidly become a landmark resource in the field, accumulating over 85 citations since its 2023 publication. By addressing a critical data gap that had long hindered progress in dexterous manipulation research — an area far less explored than simpler parallel-gripper grasping — Li's work has provided the community with a scalable foundation for training and benchmarking next-generation robotic hands. His more recent work, **PhysPart** (2025), extends his interests into 3D generative modeling, tackling the challenge of physically plausible part completion for interactable objects, with implications ranging from 3D printing to robotic simulation environment construction. Across his research, Li consistently bridges the gap between physical realism and computational modeling, producing tools and datasets that directly accelerate downstream applications in robotics and embodied AI. His growing citation record reflects a meaningful early-career impact on a rapidly evolving field.