RoboScience raises several hundred million yuan for Visics model and REX G1 humanoid

Date: 2026-11-02
Type: ARTICLE
Topics: Funding, Mobile Manipulation, Robot Foundation Models, Robot Simulation
Deck: The A2 round supports embodied-model development and logistics, retail and industrial trials after RoboMirage, Visics and REX G1 releases.
URL: https://korthosrobotics.com/research/roboscience-a2-financing-visics-rex-g1-nov-2026
Referenced profiles: RoboScience / 机器科学 (https://korthosrobotics.com/ecosystem/company/roboscience)

Image: RoboScience
RoboScience disclosed an A2 financing of several hundred million yuan on 2 November. The round was co-led by Amber Capital, Hongruida Investment and Shanghai Semiconductor Industry Investment, with participants including 360’s investment arm, Jinding Capital and other investors. According to Beijing News, proceeds will support embodied-model development, proof-of-concept work and deployment in logistics, retail and industrial settings, and technical recruitment.
Established in December 2024, RoboScience was founded by Tian Ye, previously an AI platform executive at Apple, and robotics researcher Shao Lin. Its technology stack links the RoboMirage physics-simulation engine launched in September 2025, the Visics embodied model introduced in June 2026 and the REX G1 wheeled humanoid unveiled on 17 August.
RoboScience describes Visics as a general embodied model designed to transfer manipulation skills across robot forms. A world model predicts object trajectories, while an operation model translates those trajectories into control signals for the particular robot. Through its Tencent Cloud partnership, RoboScience also offers cloud-based access to manipulation capabilities through an API, alongside its own robot hardware.
RoboScience says its video-training corpus expanded from approximately 1 million to 4 million hours between the start of 2026 and October, alongside more than 10 billion simulated motion trajectories. These company-reported data volumes support its combined video-and-simulation training approach. The company is advancing proof-of-concept programmes in retail, logistics and industry, with the new capital intended to extend model capabilities and application deployment.