① It is reported that 'Ge Wu' integrates an advanced reinforcement learning framework and multimodal motion control technology. ② Professor Ye Linqi, the open-source creator of the 'Ge Wu' platform, believes that true embodied intelligence in the future needs to meet the most basic survival needs. ③ Analysts think that the second half of 2025 will be a stage for the rapid promotion of humanoid robot commercialization.
The Star Daily reported on March 21 that this morning, the National Local Joint Construction Humanoid Robot Innovation Center, in collaboration with Shanghai University and Tsinghua University, jointly launched the embodied intelligence simulation platform 'Ge Wu', promoting the transition of humanoid robot technology from the laboratory to industrialization.
According to reports, the National Local Joint Construction Humanoid Robot Innovation Center is the first innovation center in the country focused on the core technologies of embodied intelligent robots, product development, and application ecology.$UBTECH ROBOTICS (09880.HK)$It is established by organizations such as Jingcheng Electromechanical, Xiaomi Robotics, Yizhuang Robotics, and others, with more than 70% of the research and development personnel involved.
Similar to intelligent driving, humanoid robots also require simulated environments to implement algorithm development and testing, performance evaluation, control system optimization, and other processes. According to information from the press conference, the 'Ge Wu' platform integrates advanced reinforcement learning frameworks and multimodal motion control technology, achieving technological breakthroughs in the following aspects:
One-click training: Integrating an automated training framework, the algorithm iteration cycle is shortened by more than 50%, new robot onboarding means immediate training without the need for reprogramming.
Multi-machine reuse: Compatible with 127 mainstream robot hardware, the online learning technology can simultaneously compare the configurations and motion capabilities of various robots.
One machine, multiple brains: supports rapid switching of motion intelligence models across 8 major scenarios including Industry, Service, and Medical.
One-click migration: complete seamless migration from simulation to reality through virtual and physical mapping calibration technology.
Recently, several humanoid robot manufacturers, including Yushu and Zhongqing, have promoted their products' motion control capabilities. At the launch conference, Professor Ye Linqi, the core algorithm open-sourcer of the 'Ge Wu' platform, stated that the currently showcased robots do not have continuous learning capabilities, and that future embodied intelligence needs to meet the most basic survival needs.
Xiong Youjun, general manager of the Peking Humanoid Robot Innovation Center, believes that the level of intelligence in embodied intelligence is still at an elementary stage, and the Industry urgently needs a general intelligent platform with multi-Ontology compatibility, multi-scenario adaptability, and strong generalization capabilities.
In addition to 'Ge Wu', the national and local co-built humanoid robot innovation center open fund has also been simultaneously released. This fund will set open fund topics around four major areas: platform technology, embodied intelligence, datasets, and training grounds, assessing the basic, originality, application prospects, and potential impact of topics to select some excellent projects for funding.
Meng Pengfei, chief Analyst of Open-source Securities in Machinery, recently stated that humanoid robots are the ultimate form of robotics and the most important carrier in the AI era. China possesses complete supply chain capabilities for large-scale humanoid robot manufacturing, rich application scenarios, policy support, and talent advantages, and is expected to lead in the mass production stage. From an investment perspective, 2025 is the first year of humanoid robot mass production, and it is expected that the second half of 2025 will be the rapid advancement phase for the commercialization of humanoid robots.
Editor/Rocky
Comment(1)
Reason For Report