Chinese leader in humanoid robotics is bent on integrating AI learning with stronger hardware to make humanoid robots more intelligent, robust, scalable and more commercially viable.

Beijing, October 16, 2024 -- Robot Era, a pioneer in embodied intelligence, has raised close to 300 million yuan ($42.17 million) in a Pre-Series A funding round.

Crystal Stream, Vision Plus Capital and Alibaba led the round, with participation from state-run Chengdu Orinno Capital.

Existing investors, including Lenovo Capital, Golden Resources, Jinding Capital, Wing Capital, and Tiancheng Capital, also poured money into the follow-on round. China Renaissance continues to serve as the exclusive financial advisor.

The proceeds will be used to advance the R&D breakthrough in embodied intelligence technologies and accelerate the commercialization of general-purpose humanoid robots. The funds will also go towards hiring talent and consolidating its technological lead.

Notably, founded by Chen Jianyu, an assistant professor at Tsinghua University's Institute for Interdisciplinary Information Sciences, Robot Era is the only humanoid robotics company in which Tsinghua holds equity.

Chen hailed the significance of two major technological milestones—deep reinforcement learning, as exemplified by AlphaGo, and large language models like ChatGPT—for the evolution of humanoid robots. And he envisions that these two technologies will integrate deeply with humanoid robotics and embodied intelligence.

"Embodied intelligence will require scalable, native AI algorithms trained with more diverse physical-world data," Chen states. "We believe robotics will have its 'ChatGPT moment' in the not too distant future."

In addition to Tsinghua's backing, Robot Era also receives strategic support from the Shanghai Qi Zhi Institute. The company is focused on developing native general-purpose embodied intelligence systems and has built an innovative AI-driven hardware platform.

Robot Era's native robotic model paradigm offers a new approach to AI learning that emphasizes both versatility and simplicity with strong scalability.

Its algorithms integrate multimodal inputs—such as vision, sound, and tactile data—allowing robots to process and respond to environmental stimuli in a similar fashion to humans. This cross-modal capability enhances the robots' ability to interact flexibly with their surroundings.

The company's latest full-size humanoid robot, STAR 1, uses an end-to-end learning method, relying solely on a neural network for mobility and manipulation tasks.

STAR 1 can walk and grasp objects in diverse environments, powered by its language-visual-action model. It converts language and visual inputs into precise movements. This reduces signal loss and enables rapid scalability, allowing the robot to learn new skills efficiently from minimal data. (See YouTube video link: https://www.youtube.com/watch?v=qw2y0kceAv0&t=6s)

On the hardware side, STAR 1 features a seamless integration of software and hardware. All components—from motors to the dual-leg system and dexterous hands—are optimized for rapid AI validation, boosting the efficiency of Sim2Real (simulation to reality) adaptation.

STAR 1 has increased its degrees of freedom from 20 to 55, supports more complex postures, and offers improved joint torque of 400 N·m with speeds of up to 25 rad/s. The robot has passed multiple product-level tests, meeting market standards for deployment.

Robot Era has also rolled out smaller models, including the desktop robot M1 and the wheeled robot W1, showcasing the versatility of its modular design. These robots demonstrate the company's ability to translate technological innovations into practical applications across various industries.

About Robot Era

Founded in August 2023 and incubated by Tsinghua University's Institute for Interdisciplinary Information Sciences, Robot Era is committed to the development of embodied AI and humanoid robotic technologies and products. Visit http://www.robotera.com for more information.