RealMan launches Beijing data training center around RealBOT and embodied robotics infrastructure

RealMan Robotics launched a humanoid robotics data training center in Beijing in August 2025, opening a 3,000-square-metre facility built around robot operation, real-world data collection and embodied AI development. The company said the center was divided into training and application zones, with 108 robot bodies deployed across ten application scenarios including elder care, rehabilitation, assembly lines, smart retail and catering.

The Company

RealMan Robotics positions itself as a system-level infrastructure company for embodied intelligence, with product lines across robotic arms, joint modules, wheeled robots, teleoperation networks, data services and ODM robotics services. Its public site frames the company around three connected layers: reliable robot hardware, distributed teleoperation and multimodal data collection.

The company is not only presenting a single robot body; it is building the surrounding stack needed to operate robots remotely, collect data from physical tasks and support different embodiments across service, industrial, research and commercial environments. Its core products include RM and RX robotic arms, WHJ and WHG joint modules, wheeled dual-arm robots and RealBOT, a wheeled humanoid platform.

The Product

RealBOT is RealMan’s wheeled humanoid robot. The platform combines a mobile base with a humanoid upper body, dual RX75 arms, stereo depth and LiDAR perception, autonomous navigation and AI interaction. RealMan lists the robot with 21 active degrees of freedom, 5 kg payload per arm and a 40 Ah battery for extended autonomous operation.

The robot is designed as a modular system. Its upper body includes a head, torso and arms, while the lower body uses a wheeled base that can be separated or adapted to other chassis and RealMan dual-arm platforms. The company lists 14 degrees of freedom across the two arms, three in the waist, two in the legs and two in the neck, with support for dexterous hands or grippers depending on the task.

RealBOT’s sensor stack includes five depth cameras, three fisheye cameras, one LiDAR, two IMUs, an optional head-mounted microphone array and optional tactile sensors. RealMan says the platform supports 3D reconstruction, obstacle detection, object and gesture recognition, autonomous navigation, sound source localisation and contact sensing. Its chassis and shoulder width are both listed under 465 mm, giving the robot a narrow indoor profile for corridors and dense spaces.

The Data Center

The Beijing training center gives RealBOT and RealMan’s other robot bodies a defined operating context. RealMan says the facility uses embodied dual-arm lifting robots, wheeled humanoids, drone-arm systems and quadruped platforms to generate data across structured scenarios. The company’s model is based on real-world task operation rather than only simulation, with teleoperation used both to complete tasks and collect robot trajectories.

Gasgoo reported in March 2026 that RealMan had accumulated data covering more than 1,000 tasks and tens of millions of trajectory segments through 108 robot bodies across ten application scenarios. The same report said RealMan had secured nearly RMB 500 million in financing, with proceeds directed toward product R&D, the AUTRON super factory and a global ecosystem built around hardware, data and a remote operation network.

Maturity

RealMan is more commercially grounded than many early embodied AI companies because it already sells arms, joint modules and dual-arm platforms, and Gasgoo reported more than 8,000 corporate clients globally across industrial automation, research and education, and commercial services. Its RealBOT humanoid platform is still best read as an embodied intelligence and data infrastructure product rather than a proven mass-deployed service worker. The company has disclosed hardware specifications, a training center, teleoperation infrastructure, financing and data-generation claims, but public materials do not yet show named RealBOT customer deployments, fleet-scale commercial use or customer-side performance data.

Have a robotics update Korthos should review? Send news, deployments, product releases, funding rounds, research, or media to tips@korthos.xyz or reach out on X at @agkorthos.