Wenxuan Song
Papers
4
Total Citations
44
H-Index
3
About
Wenxuan Song is a rising researcher at the intersection of robotics and computer vision, with a focus on enabling intelligent, adaptable systems for complex environments. His work centers on two key areas: quadruped robot control and novel 3D scene reconstruction techniques. Song’s major contributions include the development of **QUAR-VLA**, a vision-language-action model that integrates semantic understanding with motor control for quadruped robots, and **GeRM**, a generalist robotic model leveraging mixture-of-experts to handle diverse multi-task learning challenges. These works have garnered over 30 citations collectively, reflecting their growing impact in the field. Notably, Song has also advanced underwater perception with **WaterSplatting**, a fast 3D scene reconstruction method using Gaussian Splatting that overcomes the limitations of traditional NeRF-based approaches in modeling both geometry and the water medium. This innovation has applications ranging from naval robotics to immersive VR experiences. With a clear trajectory toward bridging high-level reasoning and low-level control, Wenxuan Song is establishing himself as a promising voice in embodied AI and autonomous systems.
Research Focus
Key Achievements
Top Papers
- 1QUAR-VLA: Vision-Language-Action Model for Quadruped Robots20 citations · 2024
- 2
- 3GeRM: A Generalist Robotic Model with Mixture-of-experts for Quadruped Robot10 citations · 2024
- 4