Ruihai Wu

Papers

Total Citations

H-Index

About

Ruihai Wu is an emerging researcher working at the intersection of 3D computer vision, robotic manipulation, and embodied AI. His work centers on two interconnected themes: geometric shape understanding and language-guided robot learning, areas that are increasingly vital for building capable home-assistant robots. Wu's contributions span both foundational perception and practical manipulation. His 2023 work on SE(3)-equivariant learning for geometric shape assembly tackles the challenging problem of reconstructing fragmented objects — such as reassembling broken bowl pieces — by leveraging mathematical symmetry properties of 3D space, earning 13 citations and marking a notable advance over prior semantic assembly approaches. His research on visual affordance learning, exemplified by DualAfford (2022), pushes the frontier of dual-gripper robotic manipulation, addressing the complexity of coordinating two robotic arms to interact with diverse 3D objects. More recently, NaturalVLM (2024) demonstrates Wu's growing interest in connecting rich natural language understanding with physical manipulation, moving beyond simplistic task commands to enable more nuanced human-robot interaction. Still early in his research career, Wu has already established a distinctive profile bridging geometric deep learning with practical robotics, positioning himself as a promising contributor to the next generation of intelligent robotic systems.