Lumos Robotics, a company specializing in embodied intelligent robots, recently completed two rounds of financing for Pre-A1 and Pre-A2, with a total amount of hundreds of millions of yuan. We have summarized the information of this round of financing and several highlights of the company:

Financing amount and investment institution

Financing rounds: Pre-A1, Pre-A2

Financing scale: The total amount reaches several hundred million yuan

Investor: Pre-A1 round led by Dinghui Investment, with Nanjing Venture Capital, Jinjing Capital, and Jinggu Shares as co investors; Pre-A2 round by Shenneng Chengyi Investment

Purpose of Funds: This round of financing will be used for the company's continued investment in embodied intelligent data and hardware fields

Basic company information

Established in September 2024

Registered address: Bao'an District, Shenzhen, Guangdong Province

Enterprise positioning: Luming Robotics has long focused on the research and sales of embodied intelligent robots and core components, and has built a full stack capability loop from real machine data acquisition, hardware ontology innovation to operating system models. Relying on our self-developed FastUMI efficient data acquisition system and high-performance robot hardware platform, we provide enterprises with embodied intelligent infrastructure covering data, hardware, and algorithms, promoting the large-scale implementation and commercial application of embodied intelligent technology in multiple fields.

The company focuses on high-value industrial scenarios such as home, logistics, and manufacturing. Its core products include LUS and MOS series humanoid robots, as well as key components such as robot joint modules and visual and tactile modules.

Luming Robot Product Matrix (Image Source/Enterprise)

Team background: Founder and CEO Yu Chao graduated from Tsinghua University and has been engaged in research in the field of robot learning algorithms since 2016. He has led the construction of Zhumi Technology's embodied robot business and participated in the development of multiple consumer grade robot products such as the robot dog "Iron Egg". CTO Cao Junliang holds a PhD in Mechanical Engineering from Shanghai Jiao Tong University and has been deeply involved in the research and development of multiple high-performance embodied robot products. Co CTO Ding Yan is a Ph.D. in artificial intelligence from the State University of New York and a former star researcher at the Shanghai AI Lab. At present, the company's R&D personnel account for over 70%, including more than ten PhDs, making it a team with profound industry experience and technical accumulation.

Technical highlights: In the past year, Luming Robotics has launched four robot products, focusing on training data and hardware ontology. The company has formed a full stack research and development capability from data acquisition, ontology design, motion control, perception algorithms to system integration.

Real machine training data is a key infrastructure for the universal operation of robots' brains, with cost, efficiency, and generalization as its core considerations. Previously, the GEN-0 large model launched by Generalist in the United States preliminarily validated the Scaling Law in the field of embodied intelligence through 270000 hours of real machine data collection, but the vast majority of its data was obtained through UMI data acquisition technology. The FastUMI efficient data acquisition system of Luming Robot is an iterative optimization and performance leap of its data acquisition scheme.

Lu Ming Robot Data Acquisition Software and Hardware System FastUMI Pro (Source/Enterprise)

Compared to traditional data acquisition technology, the FastUMI data acquisition system can increase data acquisition efficiency by three times, cost only one-fifth of traditional solutions, and accuracy can reach 1-3mm. With this core technology, Luming Robot has completed 10000 hours of real machine data accumulation and base model training, and is building a comprehensive data ecosystem around the full chain capabilities of data hardware, software, and models.

In terms of hardware ontology, Luming Robotics has launched a high-performance modular robot platform. Its self-developed high torque density integrated joint is the first robot system in the industry to achieve a dual arm load of 50 kilograms; At the same time, the cycloid joint module made of all PEEK material not only reduces weight by 40% and increases torque density by 60%, but also has low-noise operation characteristics, providing adaptive support for future diverse application scenarios of humanoid robots.

market size

The market size of humanoid robots is showing a rapid and continuous expansion trend. According to a report by the High Tech Robotics Industry Research Institute (GGII), the global market size is expected to reach 6.339 billion yuan by 2025, with China accounting for over 50%. It is expected that by 2030, the global sales of humanoid robots will approach 340000 units, and the market size is expected to exceed 64 billion yuan. Yole Group predicts that the global market for humanoid robots will reach $6 billion by 2030 and is expected to rapidly climb to $51 billion by 2035.

business progress

In terms of industrial ecological layout, Luming Robotics has established deep cooperation with leading companies such as Mitsubishi and COSCO Shipping in Japan, and introduced strategic shareholders with industrial resources such as Fosun Group, SenseTime, Dema Technology, and Jingu Co., Ltd.

We have reached cooperation agreements with leading companies such as Mitsubishi (source/enterprise)

At present, the company is gradually cooperating with well-known enterprises such as Dema Technology, a smart logistics solution and core component manufacturer, and global shipping giant COSCO Shipping on the landing of embodied intelligence in logistics and intelligent manufacturing scenarios, core component development, exploring new application scenarios, and promoting the deep integration of the robot industry chain and innovation chain, accelerating the commercialization of embodied intelligence in the industry.

Founder&CEO Reflections

Hard Krypton: Luming's Fastumi technology has accumulated 10000 hours of real machine data. At present, to what extent does large-scale and high-quality real machine training data determine the large-scale implementation of embodied intelligence?

Yu Chao: Compared to simulation training, the quality of real machine training data is the highest, as evidenced by the effectiveness of the GEN0 model trained on 270000 hours of real machine data.

Real machine training brings high fidelity dynamic information in complex physical interactions, robust modeling of real environmental noise and uncertainty, and core basis for cross scenario task generalization. It has unparalleled advantages over simulation training in reducing the transfer gap from simulation to reality (Sim2Real) and ensuring the safety and reliability of action execution. Therefore, we believe that large-scale and high-quality real machine data is the key to achieving large-scale implementation.

Currently, real robot data collection faces three core pain points: high collection cost, low efficiency, and poor cross ontology adaptability. These bottlenecks severely restrict the large-scale accumulation of high-quality training data, which in turn affects the landing speed and stability of embodied intelligence in practical scenarios.

Luming's FastUMI technology can solve the above problems, increasing data collection efficiency to three times that of traditional methods, reducing costs to 1/5, and achieving accuracy of 1-3 millimeters. This is precisely building infrastructure for efficient accumulation of such resources.

Hard Krypton: How to further optimize the closed loop of data collection and model training?

Yu Chao: Next, our team will continue to advance in the following three directions. Firstly, enrich and expand the hardware product system; The team is building a hardware product matrix that covers multiple types of grippers, force controlled/non force controlled structures, and portable forms to meet the diverse needs of customers in different industries with flexible and configurable solutions.

Secondly, a systematic upgrade of data and algorithm capabilities. In the process of data collection and processing, we are continuously optimizing the entire process, building a high standard data quality evaluation system, strengthening data credibility and processing efficiency, and providing reliable data foundation and algorithm support for the evolution of upper level intelligence capabilities.

Thirdly, develop scenario based models for the next generation of embodied intelligence. We are gradually integrating multimodal data such as tactile and force control through collaboration with cutting-edge research institutions, promoting the evolution of models towards higher dimensional perception and autonomous decision-making capabilities. In early March next year, Luming Robotics plans to release a new generation model for industrial core operation scenarios.

Triangle based on data machine scene. The overall business path of Luming is to operate high-quality whole machine products in business scenarios, accumulate real machine data, train better models, and then apply them to a wider range of scenarios. It can be said that all of our business is focused on accumulating real machine data. Combining with the commercialization process, achieving the scale of real machine data while commercializing is the main goal of Luming Robotics within 1-2 years.

Hard Krypton: The company has independently developed core hardware such as integrated joints and lightweight cycloidal modules, which have achieved breakthroughs in load and noise reduction. How will these technologies support the long-term reliable operation of robots in different scenarios such as industry and home? What is Lu Ming's strategy for balancing hardware cost and performance?

Yu Chao: When robots enter households and become consumer grade products, cost and reliability are very important considerations, which is also the focus of Luming's hardware design. In terms of cost and performance balance, we have achieved cost control while ensuring product reliability through full stack self-developed processes from ontology to core components and algorithms, laying the foundation for the large-scale implementation of robots.

At present, Luming Robot has built a hardware system that balances high load performance and family friendly characteristics.

Our integrated joint technology has achieved breakthroughs in high load performance for heavy tasks such as logistics handling in industrial scenarios. Taking Lumos robot as an example, its dual arm load capacity can reach 50 kilograms, which can stably support high-intensity work requirements.

For future home service scenarios, we have also launched a lightweight cycloidal module, which not only reduces structural weight, but also significantly reduces operating noise and improves safety performance, providing technical support for the reliable operation of robots in home environments.

Lumos Robotics, a company specializing in embodied intelligent robots, recently completed two rounds of financing for Pre-A1 and Pre-A2, with a total amount of hundreds of millions of yuan. We have summarized the information of this round of financing and several highlights of the company:

Financing amount and investment institution

Financing rounds: Pre-A1, Pre-A2

Financing scale: The total amount reaches several hundred million yuan

Investor: Pre-A1 round led by Dinghui Investment, with Nanjing Venture Capital, Jinjing Capital, and Jinggu Shares as co investors; Pre-A2 round by Shenneng Chengyi Investment

Purpose of Funds: This round of financing will be used for the company's continued investment in embodied intelligent data and hardware fields

Basic company information

Established in September 2024

Registered address: Bao'an District, Shenzhen, Guangdong Province

Enterprise positioning: Luming Robotics has long focused on the research and sales of embodied intelligent robots and core components, and has built a full stack capability loop from real machine data acquisition, hardware ontology innovation to operating system models. Relying on our self-developed FastUMI efficient data acquisition system and high-performance robot hardware platform, we provide enterprises with embodied intelligent infrastructure covering data, hardware, and algorithms, promoting the large-scale implementation and commercial application of embodied intelligent technology in multiple fields.

The company focuses on high-value industrial scenarios such as home, logistics, and manufacturing. Its core products include LUS and MOS series humanoid robots, as well as key components such as robot joint modules and visual and tactile modules.

Luming Robot Product Matrix (Image Source/Enterprise)

Team background: Founder and CEO Yu Chao graduated from Tsinghua University and has been engaged in research in the field of robot learning algorithms since 2016. He has led the construction of Zhumi Technology's embodied robot business and participated in the development of multiple consumer grade robot products such as the robot dog "Iron Egg". CTO Cao Junliang holds a PhD in Mechanical Engineering from Shanghai Jiao Tong University and has been deeply involved in the research and development of multiple high-performance embodied robot products. Co CTO Ding Yan is a Ph.D. in artificial intelligence from the State University of New York and a former star researcher at the Shanghai AI Lab. At present, the company's R&D personnel account for over 70%, including more than ten PhDs, making it a team with profound industry experience and technical accumulation.

Technical highlights: In the past year, Luming Robotics has launched four robot products, focusing on training data and hardware ontology. The company has formed a full stack research and development capability from data acquisition, ontology design, motion control, perception algorithms to system integration.

Real machine training data is a key infrastructure for the universal operation of robots' brains, with cost, efficiency, and generalization as its core considerations. Previously, the GEN-0 large model launched by Generalist in the United States preliminarily validated the Scaling Law in the field of embodied intelligence through 270000 hours of real machine data collection, but the vast majority of its data was obtained through UMI data acquisition technology. The FastUMI efficient data acquisition system of Luming Robot is an iterative optimization and performance leap of its data acquisition scheme.

Lu Ming Robot Data Acquisition Software and Hardware System FastUMI Pro (Source/Enterprise)

Compared to traditional data acquisition technology, the FastUMI data acquisition system can increase data acquisition efficiency by three times, cost only one-fifth of traditional solutions, and accuracy can reach 1-3mm. With this core technology, Luming Robot has completed 10000 hours of real machine data accumulation and base model training, and is building a comprehensive data ecosystem around the full chain capabilities of data hardware, software, and models.

In terms of hardware ontology, Luming Robotics has launched a high-performance modular robot platform. Its self-developed high torque density integrated joint is the first robot system in the industry to achieve a dual arm load of 50 kilograms; At the same time, the cycloid joint module made of all PEEK material not only reduces weight by 40% and increases torque density by 60%, but also has low-noise operation characteristics, providing adaptive support for future diverse application scenarios of humanoid robots.

market size

The market size of humanoid robots is showing a rapid and continuous expansion trend. According to a report by the High Tech Robotics Industry Research Institute (GGII), the global market size is expected to reach 6.339 billion yuan by 2025, with China accounting for over 50%. It is expected that by 2030, the global sales of humanoid robots will approach 340000 units, and the market size is expected to exceed 64 billion yuan. Yole Group predicts that the global market for humanoid robots will reach $6 billion by 2030 and is expected to rapidly climb to $51 billion by 2035.

business progress

In terms of industrial ecological layout, Luming Robotics has established deep cooperation with leading companies such as Mitsubishi and COSCO Shipping in Japan, and introduced strategic shareholders with industrial resources such as Fosun Group, SenseTime, Dema Technology, and Jingu Co., Ltd.

We have reached cooperation agreements with leading companies such as Mitsubishi (source/enterprise)

At present, the company is gradually cooperating with well-known enterprises such as Dema Technology, a smart logistics solution and core component manufacturer, and global shipping giant COSCO Shipping on the landing of embodied intelligence in logistics and intelligent manufacturing scenarios, core component development, exploring new application scenarios, and promoting the deep integration of the robot industry chain and innovation chain, accelerating the commercialization of embodied intelligence in the industry.

Founder&CEO Reflections

Hard Krypton: Luming's Fastumi technology has accumulated 10000 hours of real machine data. At present, to what extent does large-scale and high-quality real machine training data determine the large-scale implementation of embodied intelligence?

Yu Chao: Compared to simulation training, the quality of real machine training data is the highest, as evidenced by the effectiveness of the GEN0 model trained on 270000 hours of real machine data.

Real machine training brings high fidelity dynamic information in complex physical interactions, robust modeling of real environmental noise and uncertainty, and core basis for cross scenario task generalization. It has unparalleled advantages over simulation training in reducing the transfer gap from simulation to reality (Sim2Real) and ensuring the safety and reliability of action execution. Therefore, we believe that large-scale and high-quality real machine data is the key to achieving large-scale implementation.

Currently, real robot data collection faces three core pain points: high collection cost, low efficiency, and poor cross ontology adaptability. These bottlenecks severely restrict the large-scale accumulation of high-quality training data, which in turn affects the landing speed and stability of embodied intelligence in practical scenarios.

Luming's FastUMI technology can solve the above problems, increasing data collection efficiency to three times that of traditional methods, reducing costs to 1/5, and achieving accuracy of 1-3 millimeters. This is precisely building infrastructure for efficient accumulation of such resources.

Hard Krypton: How to further optimize the closed loop of data collection and model training?

Yu Chao: Next, our team will continue to advance in the following three directions. Firstly, enrich and expand the hardware product system; The team is building a hardware product matrix that covers multiple types of grippers, force controlled/non force controlled structures, and portable forms to meet the diverse needs of customers in different industries with flexible and configurable solutions.

Secondly, a systematic upgrade of data and algorithm capabilities. In the process of data collection and processing, we are continuously optimizing the entire process, building a high standard data quality evaluation system, strengthening data credibility and processing efficiency, and providing reliable data foundation and algorithm support for the evolution of upper level intelligence capabilities.

Thirdly, develop scenario based models for the next generation of embodied intelligence. We are gradually integrating multimodal data such as tactile and force control through collaboration with cutting-edge research institutions, promoting the evolution of models towards higher dimensional perception and autonomous decision-making capabilities. In early March next year, Luming Robotics plans to release a new generation model for industrial core operation scenarios.

Triangle based on data machine scene. The overall business path of Luming is to operate high-quality whole machine products in business scenarios, accumulate real machine data, train better models, and then apply them to a wider range of scenarios. It can be said that all of our business is focused on accumulating real machine data. Combining with the commercialization process, achieving the scale of real machine data while commercializing is the main goal of Luming Robotics within 1-2 years.

Hard Krypton: The company has independently developed core hardware such as integrated joints and lightweight cycloidal modules, which have achieved breakthroughs in load and noise reduction. How will these technologies support the long-term reliable operation of robots in different scenarios such as industry and home? What is Lu Ming's strategy for balancing hardware cost and performance?

Yu Chao: When robots enter households and become consumer grade products, cost and reliability are very important considerations, which is also the focus of Luming's hardware design. In terms of cost and performance balance, we have achieved cost control while ensuring product reliability through full stack self-developed processes from ontology to core components and algorithms, laying the foundation for the large-scale implementation of robots.

At present, Luming Robot has built a hardware system that balances high load performance and family friendly characteristics.

Our integrated joint technology has achieved breakthroughs in high load performance for heavy tasks such as logistics handling in industrial scenarios. Taking Lumos robot as an example, its dual arm load capacity can reach 50 kilograms, which can stably support high-intensity work requirements.

For future home service scenarios, we have also launched a lightweight cycloidal module, which not only reduces structural weight, but also significantly reduces operating noise and improves safety performance, providing technical support for the reliable operation of robots in home environments.