Nanxiang

I focus on embodied intelligence, multimodal perception, and lightweight visual segmentation.

#

I am Nanxiang, an M.Eng. student in Electronic Information at Shanghai University of Engineering Science.

My work revolves around VLA/robot manipulation, multimodal large models, and visual perception algorithms. I care about the relationship between model architecture, data feedback loops, and real-world system integration: an algorithm should work not only in metrics, but also under noisy data, execution error, and engineering constraints.

Selected work starts with LeRobot pi0.5 + SO-101, V2Net, HAFNet, and CodeLab-LLaMA2 / Xingyu MoE.

Portfolio · Publications · Competitions · Blog · GitHub