Daily AI & Robotics Wrap: Humanoid Robots Push into Service, Factory Floors, and Core AI Models Advance
The pace of development in the humanoid robotics sector continues to accelerate, with major announcements this week focusing on real-world commercial deployment, ambitious new software models for general intelligence, and critical advancements in the underlying component supply chain. Key news includes a Chinese automotive group showcasing its “human assistant” robot at a global conference, a new focus on AI models that can “think” rather than just process data, and a bold vision from a leading US humanoid firm.
AiMOGA Robot ‘Mornine’ Showcases Service Capabilities at Chery Global Innovation Conference
AiMOGA Robotics, the intelligent robotics division of the Chery Group, demonstrated its humanoid robot, Mornine, at the 2025 Chery International User Summit, illustrating a significant move toward deploying embodied intelligence in commercial service roles.
Mornine, which the company frames as a “trusted human AI assistant,” performed a fully autonomous car-purchasing reception process, including greeting visitors, explaining vehicle features, opening car doors, and collaborating with the car’s machine systems.
The demonstration highlighted the robot’s L3 (assistance-level) capability, which is a key stage in advancing from collaborative-level (L2) to more sophisticated, high-level intelligence. The L3 capability is designed not only to enhance service efficiency but also to enable continuous self-optimization of task performance across diverse application scenarios, such as complex 4S dealership environments.
- **Multimodal Interaction:** Mornine provided multilingual presentations for global visitors and utilized its vision-tactile dexterous hand and multi-expression interaction system for delicate object manipulation and real-time emotional expressions.
- **Ecosystem Integration:** The robot was featured across multiple Chery brand zones, taking on roles in product explanation and performance, which supports Chery’s “Automotive + Robot” human-machine collaboration strategy.
Canton Fair Highlights Chinese Robotics Growth and Global Market Appeal
The ongoing 138th Canton Fair in Guangzhou, China, has become a showcase for the country’s rapidly expanding robotics industry, drawing significant attention from international buyers and underscoring the shift of robots from laboratories to commercial production lines.
The fair featured a dedicated service-robot zone with 46 leading companies demonstrating a variety of cutting-edge products, including a humanoid robot capable of pouring water and a coffee-making robot that has achieved full-process automation.
Buyers from regions like the Middle East and Africa expressed strong interest, with one buyer from Saudi Arabia noting that several Chinese-made products would be welcomed in his home country. The presence of robots like a massage robot from DOBOT, which offers 15-minute on-site Chinese-style massages, illustrates the expanding scope of service robots.
“China has become a leading force in the robot industry. I am impressed by the variety and quality of these robots.” – Muntasar Albeshaier, a buyer from Saudi Arabia.
The robust growth is supported by data indicating that China’s robotics industry generated nearly 240 billion yuan (about $33.4 billion USD) in revenue in 2024. The country’s complete manufacturing chain and booming algorithm development are cited as key factors in its growing global competitiveness.
AI Researchers Target ‘Systematic Generalization’ for Next-Gen Robot Intelligence
Beyond hardware, the core AI software driving robotics is seeing a critical shift, as researchers at institutions like Google DeepMind and Stanford focus on enabling AI models to achieve “systematic generalization”—the ability to combine known elements into novel meaning.
This approach moves beyond current industry solutions that rely on jamming models with vast amounts of data. The goal is for AI agents to learn through exploration and interaction, similar to human learning, which requires orders of magnitude less data than current machine learning techniques.
- **DeepMind’s Advancements:** Google DeepMind’s recent Gemini Robotics 1.5 and E.R. 1.5 models are designed to embody reasoning capabilities to help robots “actually think.” The models have progressed from performing singular tasks, like folding paper, to more advanced functions, such as choosing weather-appropriate clothing.
- **Hardware & Software Synergy:** This software push is coupled with hardware initiatives, such as Apple Inc. working on enhancements for its MLX machine learning software and Nvidia continuing to develop its Nemotron collection of open-source AI technologies to make AI development more efficient for systems like robots.
Figure AI CEO Declares Humanoids a ‘New Species’ as Diligent Robotics Expands into Elder Care
The narrative surrounding humanoid robots reached a new level of ambition this week, as Figure AI CEO Brett Adcock told an audience at Dreamforce 2025 that his company is developing “a new species” of robot. The statement signals a high-stakes vision for humanoids as a transformative technology, with Adcock describing a future of self-replicating robots backed by major investors like OpenAI.
In a more immediate commercial deployment, the service robotics sector is expanding into new human-centric environments. Diligent Robotics, known for its Moxi hospital helper robot, announced plans to expand its operations into senior-living facilities.
This move is a direct response to growing pressures in elder-care settings, notably labor shortages and aging populations. Diligent Robotics plans to run pilots in 3-5 facilities, positioning robotics as a practical solution for resident support and logistics, marking a significant milestone for service robots beyond traditional industrial and hospital settings.
PRC Robotics Ecosystem Sees Gains in Core Components and Data Training
The People’s Republic of China (PRC) is strategically developing a vertically integrated humanoid robotics supply chain, with national and provincial policies promoting the localization of core component production to reduce foreign dependence.
While Western firms still dominate in high-precision components like reducers and certain motors, Chinese companies are making notable progress in sensor technologies. For example, local firms now account for over 70 percent of the PRC’s humanoid robot 6D torque sensor market, a crucial component for complex tasks like handling delicate objects.
The development of high-quality training data remains a critical focus. China is aggressively cultivating data training centers, including the world’s largest embodied intelligent robot data collection factory in Tianjin, which spans 130,000 square feet. This effort is designed to address the challenges posed by fragmented and low-quality data, which fundamentally differs from the data used in language models, requiring three-dimensional, multimodal, and bionic perception inputs.