Guanxing Lu

University Town of Shenzhen, Tsinghua University

Papers

Total Citations

H-Index

About

Guanxing Lu is an emerging researcher at the forefront of robotic manipulation and 3D scene representation, with a particular focus on integrating Gaussian splatting techniques into embodied AI systems. His most recognized contribution, "ManiGaussian," introduced dynamic Gaussian splatting as a framework for enabling robots to perform multi-task manipulation, garnering 33 citations shortly after its 2024 publication — a remarkable reception for work in a fast-moving field. Building on this foundation, Lu extended his research into the more complex domain of bimanual robotic manipulation with "ManiGaussian++," which proposes a hierarchical Gaussian world model to address the nuanced spatiotemporal dynamics arising from coordinated dual-arm systems. This progression reflects a deliberate and ambitious research trajectory: from single-arm generalization to the considerably more challenging problem of multi-body collaboration. Lu's work sits at a compelling intersection of computer vision, world modeling, and robot learning, contributing tools that help robots better understand and interact with their environments. For students and researchers exploring embodied intelligence and neural scene representations, Lu's papers represent some of the most promising recent directions in grounding vision-based world models within real robotic systems.