The future embodied robots should not just be single process robotic arms or cleaning machines, but intelligent assistants that can switch multiple skills in different scenarios. We believe that the B-end first and C-end landing of embodied robots is a feasible commercialization path, and the generalization ability of large models is the core support for promoting this cross scenario landing. ”Recently, Wen Airong, Vice President of Beijing Galaxy General Robot Co., Ltd. (hereinafter referred to as "Galaxy General"), stated in an interview with Economic Reference Daily. In recent years, embodied intelligent robots have become an important frontier direction for the implementation of artificial intelligence applications, and their ability to "generalize across scenarios" has been highly anticipated by the industry. At the capital level, the financing amount of the global robotics track is also significantly increasing. According to data released by IT Orange, the primary market financing amount in the robotics field has reached 38.624 billion yuan in the first eight months of this year, which is 1.8 times the total financing amount for the whole year of 2024. Among them, embodied robots are considered to have the potential to become the new generation of "super terminals" after smartphones due to their "human centered" universal carrier. As one of the national high-quality dataset series achievements, Galaxy General Motors released a high-quality synthetic dataset driven embodied large model and open-source synthetic dataset in late August. Its star product embodied large model "Galbot" has been normalized in retail scenarios. In late June, Galaxy General Motors completed a new round of financing of 1.1 billion yuan led by CATL and Puquan Capital. Since its establishment only two years ago, Galaxy General Motors has accumulated over 2.4 billion yuan in financing. The impressive financing results, remarkable technological achievements, and constantly enriching commercial scenarios complement each other, outlining the strong development trajectory of Galaxy General Motors in the field of embodied large models. Unlike traditional specialized robots that only perform a single task, Galaxy General Motors hopes to enable robots to shuttle between different scenarios. ”Wen Airong said. In order to achieve this goal, Galaxy General Motors has not only developed robot hardware, but also developed its own large-scale models. In the retail scene, empowered by the Galaxy Universal end-to-end embodied large model "Grocery VLA", robots can flexibly grasp tightly placed beverages, bottles, bags, and even different forms of products hanging on hooks. Wen Airong said that the full chain closed-loop capability of "perception understanding decision execution" is a key technology for universal robots to achieve cross scenario generalization. It is reported that at the first World Humanoid Robot Games held in August, Galaxy Universal Robot won the championship in the medical sorting scene competition. In addition, Wen Airong also mentioned that in the healthcare scene, embodied intelligent robots have great application prospects in physical examination guidance, drug distribution and delivery, night duty, patient information import, surgical rehabilitation and other aspects. Universal robots do not replace humans, but complement them. Robots can replace humans in repetitive, tedious, and even dangerous tasks, allowing people to devote more energy to more creative and valuable tasks. ”Wen Airong said that in the retail business, Galaxy General Robot has achieved efficient operation and rapid store expansion for partners, and has also brought consumers a new shopping experience; In the nighttime retail scene, robots are autonomously on duty, forming human-machine complementarity, significantly improving efficiency and reducing labor costs; In the context of healthcare, robots can also serve as communication partners for the elderly, bringing unexpected emotional comfort. But Wen Airong admitted that different family environments have huge differences and complex and diverse tasks. In order for robots to truly enter the home and provide flexible services like nannies, it is necessary to collect data in more scenarios, continuously train and iterate models. (New Society)