Lingfeng Zhang
Lingfeng Zhang is a first-year Ph.D. student at Tsinghua University's Shenzhen International Graduate School, supervised by Prof. Wenbo Ding and Prof. Xiaojun Liang, and also serves as an intern at Xiaomi EV. His research focuses on embodied intelligence, embodied navigation, and vision-language-action models. He is the lead author of OneVLA, a unified framework that handles both navigation and manipulation tasks within a single model architecture, and has published work at venues including ACM MM, ACL, and ICRA.
“The central achievement of OneVLA is eliminating the architectural split between navigation and manipulation models. Every prior approach — including the best cross-task competitor — required separate action heads or fine-tuned model variants for each task type. OneVLA runs both from one set of weights.”
Source→“Lingfeng Zhang — Tsinghua University, First Author. Affiliated with Tsinghua and Pengcheng Laboratory.”
Source→AI-extracted from podcast / newsletter / paper summaries. May contain errors.