R

RoboScience

Robotics · Shenzhen, China · Founded 2024

RoboScience, also known as Beijing Jike Future Technology Co., Ltd, develops general embodied intelligence for robots. The company was established in December 2024 and is based in Beijing, China. The founders include CEO Tian Ye, formerly a machine learning tech lead at Apple, and Chief Scientist Shao Lin, an assistant professor at the National University of Singapore. Both Tian Ye and Shao Lin are alumni of the Stanford AI Lab. The company's core business is the research and development of cross-embodiment general-purpose embodied intelligence, which enables robots to perform universal manipulation tasks across different hardware, objects, and scenarios. RoboScience develops wheeled quadruped robots and integrated model architectures that feature full-stack multimodal perception, including centimeter-level environmental mapping and millisecond-level path planning. In 2024, the company launched its Manipulation Foundation Model. More recently, it introduced an end-to-end VLOA (Vision-Language-Object-Action) large model that combines an embodied world model with an embodied manipulation foundation model. This system aims to create a general intelligence framework adaptable to any task, object, and robot. RoboScience targets a wide range of applications in sectors such as manufacturing, logistics, retail, and households. The business model involves providing embodied intelligence solutions and licensing its proprietary AI models to robotics manufacturers and automation enterprises. The company has secured significant funding, including a seed round in April 2025 and an angel round of nearly CNY 200 million (USD 27.9 million) in July 2025, led by JD.Com. A Pre-Series A funding round in February 2026 raised several hundred million RMB, with investors including Puhua Capital, Fortune Venture Capital, and returning shareholder China Merchants Venture. Keywords: Embodied intelligence, general-purpose robotics, cross-embodiment AI, manipulation foundation model, wheeled quadruped robots, multimodal perception, VLOA model, robotic learning framework, self-supervised training, vision-language-action, industrial automation, logistics robots, retail robotics, household robots, Tian Ye, Shao Lin