Daily AI and Robotics Wrap: Humanoid Advancements and Commercial Deployment Push
Richtech Robotics Unveils Advanced Humanoid Robot Dex at CES 2026
Richtech Robotics Inc. is set to showcase its enhanced humanoid robot, Dex, at the upcoming CES 2026 in Las Vegas. The company’s presentation will highlight Dex’s operational capabilities in dynamic, real-world commercial and industrial settings, signaling a move from lab prototypes toward practical deployment.
The Dex robot is powered by the NVIDIA Jetson Thor platform, which provides the necessary computing power for real-time reasoning and high-precision execution of complex tasks. The integration of this advanced hardware is a key factor in the robot’s design for operating in unpredictable environments.
During the CES demonstration, Richtech plans to show Dex performing a variety of service-oriented and logistics tasks, including acting as a barista, handling food delivery, and conducting visual monitoring. The robot is designed for a four-hour operation time on a single charge.
The focus on showcasing Dex in commercial roles underscores the industry trend of targeting productivity and efficiency gains in sectors like hospitality, logistics, and manufacturing. Richtech’s strategy centers on delivering reliable, AI-driven automation for enterprise customers.
China Opens Robotics School for Humanoid Training to Accelerate Real-World AI
In a major development for the global robotics ecosystem, China has opened the second phase of the Beijing Humanoid Robot Training Center, a specialized facility designed to train humanoid robots for diverse real-world applications. The center is a key part of China’s national strategy to accelerate the development and deployment of intelligent humanoids.
The two-story complex is structured to replicate various real-life environments, including production lines, smart homes, and public spaces. This setting allows robots to undergo practical, hands-on training for tasks ranging from sorting parcels and packing goods to cooking and organizing domestic spaces.
The primary goal of the center is to overcome the critical challenge of data scarcity in robotics. According to the head of the training center, advanced AI models and humanoid robots struggle to generalize skills across different environments without vast amounts of standardized, high-quality data. The facility is expected to generate millions of data points annually.
Robots at the school, such as the 165 cm tall Kuafu developed by Leju Robotics, are paired with two human instructors in a process likened to a child’s repeated training to learn to walk. The focus areas for training include industrial production, smart home applications, elderly care services, and scenarios leveraging 5G integration.
Humanoid Robots: Novelty Acts Today, But Investment Surging Despite Technical Hurdles
An industry analysis suggests that while investment in humanoid robotics is surging globally, the technology remains largely in the “marketing and experimentation” phase, performing more as novelty acts than commercially viable workers at scale. This viewpoint highlights the persistent gap between impressive promotional videos and sustained, reliable performance in complex, real-world environments.
Experts point to two major barriers to widespread commercial deployment: technical limitations, particularly in manual dexterity, and the high cost compared to human labor. Dexterity is often described as the “last frontier,” with current models still performing simple tasks like folding laundry slowly and with limited precision.
The ROI has to be ‘the humanoid robot has to do 100 percent of what the human can do for no more money than a human costs.’
The challenge of data availability is also critical, as machine learning models require vast quantities of real-world sensory data to achieve human-equivalent performance. The process of gathering this information is expected to take significant time and effort.
Despite the current limitations, interest and investment are at an all-time high, with approximately 50 companies worldwide credibly developing humanoids, including about 20 in China and 15 in North America. This surge in capital suggests a strong belief that the technical challenges will be overcome, leading to a decades-long transition to scaled commercial use.
Hyundai Confirms Boston Dynamics’ All-Electric Atlas Debut at CES 2026
Hyundai Motor Group has confirmed that the all-electric version of Boston Dynamics’ Atlas humanoid robot will make its first live public demonstration at CES 2026. The debut is a centerpiece of Hyundai’s broader strategy to position AI robotics at the core of its future growth and outlines a blueprint for building an integrated AI robotics ecosystem.
The new, all-electric Atlas is a significant evolution from its hydraulic predecessor, featuring enhanced strength, dexterity, and a range of motion that reportedly exceeds human capabilities. It is designed for commercial deployment, with pilot demonstrations already underway at a Hyundai Motor Group EV plant in Georgia for automotive manufacturing applications.
Hyundai’s plans for its robotics division, which includes Boston Dynamics, involve managing the entire value chain from AI robotics development and training to operations. This integration is set to span robot components, logistics, and software development, demonstrating a commitment to full-stack control over the technology.
- **Robot:** All-electric Atlas humanoid.
- **Event:** First live public demonstration at CES 2026.
- **Context:** Part of Hyundai’s group-wide AI robotics strategy.
- **Deployment:** Pilot trials already in progress at a Hyundai EV plant.
New AI Chips and Chinese Firms Drive Commercial Humanoid Deployments
The commercialization of humanoid robots is being driven by advancements in specialized AI processing chips and the rapid scaling efforts of Chinese robotics firms. The convergence of these two factors is pushing humanoids into real-world industrial applications at an accelerating pace.
Chinese companies are leading the charge in early commercial deployments. Fourier Intelligence, for example, is claiming its second-generation GR-2 bipedal robot is the first commercial humanoid being used by car makers. Similarly, UBtech has reportedly received 500 pre-orders for its humanoid prototypes from Chinese car manufacturers, while Shanghai Kepler Robotics is working with 50 customers on real-world scenarios in manufacturing, warehousing, and logistics.
The underlying intelligence for these robots is increasingly moving to the “edge” with new AI accelerator chips. Companies like Israeli firm Hailo and others are launching edge AI processors designed to run generative AI models, including large language models (LLMs) and vision-language models (VLMs), directly on the device. This on-device processing capability is crucial for providing the real-time reasoning and adaptation necessary for robots to operate autonomously in dynamic, unpredictable environments.
The development of neuromorphic edge AI technology, such as BrainChip’s Akida platform, further supports on-device LLMs, enabling real-time and private Generative AI without relying on cloud infrastructure. This focus on low-power, high-performance edge computing is a key technological enabler for scaling humanoid deployment.
2025 Review: Humanoid Robots Move from Controlled Demos to Factory Trials
A review of technology developments for the past year highlights that 2025 marked a definitive shift for humanoid robots, moving them beyond heavily controlled demonstrations and into actual factory and warehouse trials. This transition is seen as a key step toward the vision of general-purpose robots in homes, factories, and service roles.
Tesla’s Optimus, for instance, made its first real-world appearances, demonstrating its ability to perform actual tasks rather than just choreographed movements. This progress, alongside the rapid entry of Chinese competitors like Unitree, has intensified the global race in humanoid robotics.
Despite the impressive progress in mobility and dexterity, technical challenges remain a primary focus for the industry. Ongoing issues with battery life, complex decision-making in unstructured environments, and the high initial cost continue to be major hurdles. However, the pace of technological advancement suggests a decade of rapid growth and deployment ahead for the sector.