RoboScience Closes 1 Billion RMB Series A
RoboScience, a general embodied AI company, announced the completion of a landmark 1 billion RMB Series A funding round on May 6, 2026. The round was arranged by Huaxing Capital, with participation from multiple well-known domestic and international industrial giants and top-tier financial institutions.
Funding Allocation
The raised capital will be directed toward two key areas:
- VLOA Large Model Development: Continuing to deepen the core Vision-Language-Object-Action (VLOA) large model technology to build a more powerful general embodied AI 'brain'
- Robot Mass Production: Advancing the engineering and mass production of self-developed robot bodies, providing scalable physical carriers for technology deployment
Technical Breakthroughs
RoboScience's VLOA model focuses on the essence of robot-physical world interaction. The model's core innovation lies in its architecture and data strategy:
- Object-centric world model: Building cognition around object-level states, 3D trajectories, contact relationships, and physical causal changes
- Massive data accumulation: Through a fully automated process, the team has accumulated millions of hours of object-centric high-dimensional multimodal operation data
- Simulation data loop: Through self-developed multimodal physics engines, hundreds of billions of high-quality operation trajectory data have been generated
The company's D(R,O)Grasp framework won the ICRA 2025 Best Paper Award for robot manipulation and locomotion, while the latest T(R,O)Grasp achieved 94.83% average success rate in cross-agent grasping tasks.
Market Context
According to statistics, China's embodied AI sector has seen disclosed financing exceeding 345 billion RMB in 2026 so far. Capital is flowing from complete machines to upstream core components like joint modules, sensors, and dexterous hands.
