From Vision + Multimodal Data Collection
to Embodied Deployment

🎓 Our Team

Our two co-founders come from NUAA (Nanjing University of Aeronautics and Astronautics) and SJTU (Shanghai Jiao Tong University), both long-time SLAM researchers in multi-sensor fusion, hardware-level synchronization and spatiotemporal alignment.

We know the first barrier for embodied VLA from pretraining to deployment is not algorithms — it's high-quality, multimodal, time-synchronized first-person data.

• NUAA · Co-founder — SLAM & multi-sensor fusion, system architecture
• SJTU · Co-founder — visual-inertial odometry & hardware sync, capture pipeline
• Shared belief — serve scene owners, help VLA teams deploy for real

💡 The Industry Problem We See

Every VLA startup struggles with funding: angel rounds of a few million RMB. To amass 1,000 EGO devices and 1 million hours of vertical pretraining data, buying terminals at ¥5,000–10,000 each simply doesn't add up —

The money ends up with speculators selling overpriced hardware at the peak and leaving, while real data teams struggle.

• Pain — expensive terminals, closed sync solutions, incompatible formats
• Malpractice — overpricing, hit-and-run vendors, no after-sales
• Our stance — long-term builders, not speculators

🚀 Our Advantages

Our hardware-software-algorithm DNA gives every EGO capture terminal ground-truth-level data capability:

• Annotation-free pretraining data — body pose / hand joint pose captured at sub-millimeter ground truth accuracy, meaning no online annotation needed
• Easy HW integration of any third-party gloves/EMG bands — EGO-Hub provides hardware external-trigger synchronized sampling for any third-party glove / wristband sensor