Sorting materials, handling material boxes, precision assembly... Recently, reporters walked into Geely Automobile's Jike 5G smart factory located in Ningbo, Zhejiang. They saw dozens of humanoid robots - Walker S1, distributed in multiple complex scenarios such as the final assembly workshop, instrument area, quality inspection area, and door assembly area, successfully achieving multi task collaborative operations such as sorting, handling, and precision assembly. The reporter learned from Shenzhen Ubiquitous Technology Co., Ltd. (referred to as "Ubiquitous"), which develops and manufactures the Walker S1, that this is the first collaborative training of multiple, multi scenario, and multi task humanoid robots carried out by the company. The company hopes to explore the establishment of a universal humanoid robot group operation solution for multi task industrial scenarios, and promote the evolution of humanoid robots from single machine autonomy to group intelligence. In recent years, with the rapid development of artificial intelligence technology, many technology companies at home and abroad have introduced humanoid robots into the automotive production field. For example, the Apollo humanoid robot produced by Apptronik in the United States has entered the production line of Mercedes Benz, while the Figure01 humanoid robot has entered BMW's factory in Spartanburg, USA. Currently, leading companies in the industry such as BYD, Foxconn, SF Express, and BAIC New Energy have partnered with Ubiquitous. Ubiquitous's independently developed industrial humanoid robot Walker S series has successfully entered multiple car factories worldwide and completed the first phase of single machine autonomous intelligent training. Our innovative humanoid robot BrainNet architecture and Intelligent Networked Hub (IoH) architecture provide a feasible technological path for collaborative software and hardware in swarm intelligence According to Jiao Jichao, Vice President and Dean of the Research Institute of Ubiquitous, Ubiquitous has initiated an internal research project on embodied intelligence to explore the evolution of high-order embodied intelligence through intensive interaction between group behavior and external physical environment. This project is based on the basic concept of a super brain and an intelligent cerebellum, and maximizes the functions of the brain in the cerebellum, resulting in the Ubiquitous Walker S1 equipped with a swarm brain network. The technological breakthrough of swarm intelligence is a necessary path for the large-scale deployment of humanoid robots in manufacturing scenarios. Jiao Jichao introduced that in the Youbixuan swarm brain network, the super brain is based on a multimodal embodied reasoning large model, performs multi-scale model collaboration and cloud edge cognitive topology, responsible for advanced semantic understanding, reasoning decision-making, and potential anomaly monitoring. It focuses on breaking through intelligent hybrid decision-making technology and achieving high-dimensional decision-making for complex production line level tasks; The intelligent cerebellum is a perception+operation+motion end side small model based on the Transformer architecture. It develops cross domain fusion perception technology and multi machine collaborative control technology to achieve efficient and accurate decision control, while supporting multi machine parallel distributed learning to accelerate the generation and transfer of humanoid robot capabilities. It is worth noting that the super brain mentioned above also utilizes the humanoid robot multimodal embodied reasoning large model developed by Ubiquitous based on DeepSeek-R1 deep reasoning technology. The Walker S1, equipped with this embodied reasoning model, can build a billion level high-quality industrial dataset by accumulating training tasks on the front lines of multiple car factories. This high-quality dataset accumulated in real scenarios is currently the most lacking in the iterative evolution of humanoid robots, "said Jiao Jichao. It is reported that it is precisely with such a swarm brain network architecture, high-quality data accumulation, robot skill iteration, and supplemented by the Internet of Humanoids (IoH), which expands the scope of tasks autonomously executed by a single humanoid robot to meet the flexible requirements of multiple robots working together at the production line level, that Walker S1 demonstrates multi-dimensional and multi scenario swarm intelligence capabilities in the Extreme Krypton 5G smart factory. Humanoid robots are designed to perform tasks ranging from simple logistics to complex assembly, alleviating the pressure of skilled worker shortages and improving production quality. However, humanoid robots still face significant challenges in terms of technological maturity, cost control, safety, and human-machine collaboration. Some experts remain skeptical about the entry of humanoid robots into the industrial manufacturing field. The founder of Boston Dynamics, Robert, has made it clear that he is not optimistic about the prospects of this humanoid robot in industrial manufacturing. He believes that automobile manufacturing is the advantageous field of traditional robot companies, and robots such as welding, assembly, spraying, and handling have already played important roles on the production line. In contrast, the role of humanoid robots on the production line is limited and can only be considered as "mascots"; Abderrahmane Kheddar, an academician of the French National Institute of Technology and a professor at the University of Montpellier, believes that although robots can improve efficiency and production quality in manufacturing, the advantages of humanoid robots have not yet been fully demonstrated. Although internationally renowned investment firm Goldman Sachs predicts that the global market size of humanoid robots will reach $38 billion by 2035, humanoid robots are currently unable to truly work like humans. Further research and exploration are needed for future development. Jiao Jichao stated that in complex scenarios, swarm intelligence emerges global intelligence beyond a single ability through dynamic interaction and collaboration of distributed individuals, which can help humanoid robots make efficient decisions and adaptively optimize, breaking through large-scale commercial applications across scenarios and tasks. This breakthrough not only provides a reference paradigm for the large-scale application of humanoid robots in intelligent manufacturing, but also promotes the transition of new industrialization from "semi flexible manufacturing" to "fully flexible intelligent manufacturing" paradigm. (New Society)