GENISOM AI steps into the global robotics spotlight at ICRA 2026 with real-world-ready robots - Robotics & Automation News
Its dual hot-swappable batteries support continuous operation, with battery life of up to five hours. These features make it practical for fieldwork where reliability, endurance, and adaptability are important.
Key takeaways
Humanoid robot activity is accelerating worldwide. In Beijing, Lingyi iTech’s new factory has already built 300 units just weeks after opening and aims to reach 10,000 this year, with a long‑term goal of 500,000 annually by 2030, a scale that could halve the current $30,000 price tag. China is also assigning digital ID numbers to every humanoid to track safety and lifecycle data, while a government showroom is displaying models such as the 199,000‑yuan Booster T1 and the 350,000‑yuan R1 Pro, with cumulative orders topping 30 million yuan as of late May. Demand, however, still lags capacity: Omdia reports that 13,000 humanoids shipped in 2025, and Morgan Stanley expects Chinese sales to double to about 28,000 units in 2026, with annual shipments potentially exceeding one million by the early 2030s. Companies such as Matrix Robotics, Unitree and AGIBOT are scaling production, with Unitree delivering over 5,000 units in 2025 and Matrix targeting 5,000 deliveries this year after receiving roughly 1,000 orders from hotels and coffee chains. In the West, Standard Bots secured a $1 billion valuation after a $200 million Series C, while Figure announced rapid ramp‑up of its humanoid line and a logistics partnership with Catalyst Brands. The UK‑based Humanoid firm signed a proof‑of‑concept deal with Bosch to move toward mass production, and Boston Dynamics pledged to deliver up to 25,000 Atlas units to customers such as Hyundai, expanding its capacity to 30,000 robots per year by 2028. New hardware breakthroughs include Rotaku’s Domo platform for developers priced at $2,999, Sharpa’s Wave tactile hands integrated into Unitree’s H2 Plus reference design on Nvidia’s Isaac GR00T framework, and MIT’s ultrasound wristband that captures hand‑muscle motion to generate training data for dexterous manipulation. Agile Robots showcased force‑control and humanoid technologies at the Robot Technology Japan event, highlighting its AgileONE humanoid and recent acquisition of thyssenkrupp Automation Engineering. Across these developments, the industry is moving from lab demos toward scaled manufacturing, tighter software integration, and real‑world commercial deployments.
Its dual hot-swappable batteries support continuous operation, with battery life of up to five hours. These features make it practical for fieldwork where reliability, endurance, and adaptability are important.
In power grid inspections, the GENISOM M1 can support teams working around substations, power lines, and other infrastructure.
In industrial environments, it can assist with routine inspection tasks in areas that may be difficult, dangerous, or time-consuming for human workers to access.
For emergency response, the robot’s mobility and payload capacity can help carry sensors, communication tools, or other equipment into challenging locations. ### Real-World Applications Across Industry and Infrastructure
Technology remains an important part of GENISOM AI’s story, but the focus is on practical value rather than complex algorithmic explanations.
The MATRiX simulation platform helps robots train and validate capabilities in simulation before entering real-world environments. This allows developers to test movement, navigation, and operational behavior before physical deployment.
RoamerX, GENISOM AI’s autonomous navigation system, supports mapping, path planning, and obstacle avoidance. In simple terms, it helps robots understand where they are, plan where they need to go, and respond to obstacles along the way. The GENISOM L1-W features a hybrid wheeled-legged mobility system, combining the terrain adaptability of legged robots with the movement efficiency of wheels. This makes it useful in environments where robots may need to move across both flat surfaces and more complex terrain.
The University of Manchester team won first place in the IROS 2025 Quadruped Robot Challenge using GENISOM L1 EDU, showing the platform’s effectiveness in a demanding robotics environment.
GENISOM L1 EDU supports SLAM, 3D mapping, and open-source development, making it relevant for universities, laboratories, and developers exploring robotics applications. For education and research users, this provides a platform for testing ideas, developing software, and studying real-world robotic behavior.