Embodied AI FRONTIER
- Your Embodied AI Intelligence Hub -
/ Daily / RealMan Builds 150-Robot Data Training Ground in Changzhou to Tackle Embodied AI Data Challenges
Industry

RealMan Builds 150-Robot Data Training Ground in Changzhou to Tackle Embodied AI Data Challenges

Ji Haifeng, solution director at RealMan Intelligent Technology, provided a detailed breakdown of the current state of embodied AI data training at the 2026 Embodied Perception Fusion and Multimodal Large Model Innovation Symposium. He highlighted four recurring pain points in the industry that are frequently mentioned.

The first is the scarcity of high-quality real-machine data. The second is the high cost of collecting such data, with a single robot costing up to hundreds of thousands of yuan. The third is the gap between simulation environments and the real physical world. The fourth is the low level of industry standardization, with national standards still missing.

RealMan has taken actions in these areas, including building data collection facilities, participating in standard-setting, and developing data collection methodologies.

In May 2026, RealMan completed a data training ground in Changzhou equipped with 150 robots, open to the public. The facility is divided into two main functional areas: a basic action training zone for grasping, picking, and placing, similar to how a child learns basic motions; and a scenario application zone that sets up diverse business scenarios like home, industrial, and retail for robots to perform practical tasks.

Ji noted that real-machine data collection is not a simple one-time execution. It requires generalization across multiple dimensions, including color, objects, and actions. This process is akin to a child learning to walk and grasp objects, requiring repeated practice. The data collection follows the MCAP standard, which sets specifications for six key characteristics: time synchronization, data integrity, data accuracy, data consistency, data usability, and the difficulty of collection actions.

RealMan has participated in the drafting of multiple data-related standards and is also promoting the formulation of corresponding national standards. This aligns with the industry's need for standardization.

Data volume is a significant but often overlooked issue in training ground operations. Ji revealed that a single robot can generate 2GB of raw data per minute, and even after compression, the data volume still reaches 0.8GB. This massive volume dictates that the training ground's underlying architecture must treat storage and networking as infrastructure layers. Storage modes are divided into local and cloud storage. Cloud storage requires strict network conditions, including dedicated lines, to upload data to the cloud. Its advantage is convenient delivery, allowing direct migration of data to customers.

The MCAP data format contains various types of data. Image data is divided into compressed and raw images. In addition, there are camera calibration data, joint posture data, velocity data, pose data, end-effector-related data, and six-dimensional force data. Ji particularly emphasized time synchronization. The robot system has multiple data streams from cameras, robotic arms, and posture actions. It is essential to ensure that all data types start recording at the same time point and within the same time window. A model R&D company that purchases RealMan's data has required time deviation to be controlled within 5 milliseconds.

Remote operation is a compromise solution. RealMan found that the model of data collection relying on robots within closed venues has shown certain drawbacks, similar to the limitations of keeping a child indoors. The company advocates deploying robots in real scenarios to collect operational data. The logic of remote operation is that operators remotely teleoperate robots to complete tasks at factory sites. While performing the task, data collection is also completed. The business operation is the primary goal, and data collection is an incidental output that does not interfere with actual business operations. Ji divides the development cycle of this model into three stages. The first stage is remote operation. The second stage achieves a data closed loop, gradually reducing and even eliminating human intervention. The final stage realizes autonomous evolution of robots.

RealMan's robotic arms have achieved 50,000 hours of failure-free operation, and the company is aiming for 80,000 hours this year. Starting with ultra-light humanoid robotic arms, the company has self-developed complete robots, mainly wheeled robots, and has no plans for legged robots. Deployment cases include providing robot bodies and supporting data to a domestic research institute, providing data support to Ant Lingbo, and exporting data to a Japanese research institute. Ji has a judgment on cost: at present, a single robot is priced at hundreds of thousands of yuan. To achieve home popularity and large-scale factory applications, the robot cost needs to drop to the tens of thousands of yuan level. Whether this cost target can be achieved is directly related to whether the data training ground can continuously produce high-quality data and improve collection efficiency.

✓ Verified 2026-09-16
Flash Briefs Live
Today · Sep 16 Wed 19
04:28
Capital 96
Zhejiang University-affiliated Tuorui Intelligence Completes Tens of Millions Yuan First Round
04:19
Industry 94
Robot Tug-of-War, Obstacle Course, Jumping: 2026 Embodied AI Challenge at Digital Trade Expo
04:08
Industry 94
CPIC P&C Issues China's First Embodied AI Component Insurance for Dexterous Hands
04:08
Industry DEEP Robotics 97
DEEP Robotics shows off robot barista in embodied AI push
04:03
Tech 96
Shangpin Home Collection Open-Sources WorldSimReady-Home Dataset for Robot Training
04:00
Industry 94
RealMan Builds 150-Robot Data Training Ground in Changzhou to Tackle Embodied AI Data Challenges
03:52
Tech 96
ZhiXing Robot Takes On Lacquer Fan Craft at Hong Kong Mid-Autumn Lantern Fair
03:42
Tech 96
Israel's Physical AI Push Needs to Account for the Invisible Electromagnetic World
03:35
Industry 98
Unitree Unveils H2 Humanoid: 180cm, 70kg, Ballet and Kung Fu Capable
03:28
Industry 96
Robot Era: Half Overseas, Nine of Top 10 Tech Firms as Clients
03:20
Industry 96
Live Science Poll: 42% Would Let Humanoid Robots Handle Chores, 36% Distrust
03:14
Industry 96
40+ AI Firms Demo Real Work at Bund; UDI Robot Jumps 150% in HK Debut
03:06
Industry 96
How Much Water Is in Robotics' Celebrated Order Announcements?
02:58
Industry 96
China's 70+ Robot Training Schools Struggle to Sustain Without Data Sales
02:54
Capital 96
Humanoid Robot IPO Reviews Tighten as Unprofitable Firms Face Profitability Test
02:43
Industry 96
China's Brain-Computer Interface Firms Top 200, but Researcher Says It's Not Yet a Rigid Need
02:38
Industry 96
Agility's Digit 5 Steps Back from Humans, a Key Step for Humanoids in Factories
02:00
Industry 98
Shenzhen's Xingcan Smart Brings Autonomous Driving to Elder-Care Wheelchair at APEC
00:03
Tech 98
15-Year-Old Builds Bionic Turtle Robot That Swims Without Propellers
Yesterday · Sep 15 Tue 37
23:57
Industry 98
Humanoid Robot OEMs Struggle as Dexterous Hand Profits Shift Upstream
23:40
Policy 98
Hong Kong Chief Executive John Lee Positions AI as Key Industry in First Five-Year Plan
22:38
Tech 98
Hefei Team Builds Forestry Inspection Robot for All-Weather Fire Detection
21:24
Capital 98
Hong Kong Stocks Open Higher; Robot Concept Stocks Surge, Lithium Battery Leader Falls
15:56
Industry Agility 99
Agility CEO Peggy Johnson talks AI-driven robot skills on Bloomberg TV
12:30
Industry Agility 98
Agility CBO: Home Humanoids Need Cooperative Safety First
11:22
Industry 98
Agility's Digit 5 May Be the First Truly Safe Humanoid Robot Worker
10:32
Industry Boston Dynamics 99
Boston Dynamics updates Spot and Orbit for agentic factory workflows
09:50
Capital 99
Hangcha Group Plans 2.259 Billion Yuan Convertible Bonds for Forklift Robot Expansion
09:19
Industry MOONS' INDUSTRIES-Motor and Drive Manufacturer 99
MOONS' unveils LiDAR motor solution for intelligent driving
09:00
Industry 99
Shangluo Electronics Subsidiary's Robot Joint Modules and Motor Drives in Sampling Stage
08:58
Industry 99
InOrbit.AI Releases OpenRobOps, an ISO 21423 Reference Implementation
08:21
Industry Hesai Technology 100
Hesai unveils MT series lidar for high-precision 3D mapping
08:21
Capital 99
Hangcha Group to Raise Up to 2.259 Billion Yuan for Forklift Robot Projects
08:17
Capital 99
CATL Subsidiary to Invest 1.12 Billion Yuan in AI and Embodied Intelligence Fund
08:00
Industry UBTECH Robotics 100
UBTECH humanoid robot factory rolls out 1 unit per 10 minutes
07:09
Industry Agility 99
Agility Robotics unveils Digit 5 humanoid for barrier-free work
07:00
Industry Agility 100
Agility Robotics unveils Digit 5 humanoid robot for industrial work
07:00
Industry lumos robotics 99
Lumos Robotics launches NexCore robot skill workflow platform
06:55
Industry 99
Agility Unveils Digit 5 Humanoid With New Legs, Batteries and Safety Upgrades
06:07
Industry LimX Dynamics 99
LimX podcast: how far are robots from real work?
04:07
Policy 93
Trump Calls Jensen Huang to Say AI Danger Theory Is a Scam
04:06
Industry HONOR 94
HONOR Robot Phone Opens Agent Skills Interface to Developers
04:05
Industry HONOR 93
HONOR Robot Phone Adds YOYO Robot Mode on MagicOS
04:02
Industry 93
Sichuan Bets on Exoskeleton Robots: Wellness and Tourism Lead, Medical Rehab Core
03:59
Industry HONOR 95
HONOR Cuts Robot Phone Gimbal Size 65%, Boosts Strength 200%
03:59
Industry HONOR 94
HONOR Robot Phone Packs 4-DoF Gimbal, 100+ Patents
03:44
Industry 93
Oke Precision Says Whirlwind Milling Tools Can Process Humanoid Robot Lead Screws
03:41
Industry 93
Companion Robots Sell Nearly 300,000 Units: Embodied AI's First Real Market Is the Living Room
03:36
Industry 93
Shendu Jizhi Brings Embodied AI to Classrooms with PrimeLite Robot
03:22
Industry 94
Blue Insect Embodied launches modular Mantis Standard humanoid robot from RMB 9,800
03:19
Industry 96
Lei Jun Visits Unitree, Lightly Kicks Humanoid Robot to Test Balance
03:07
Industry 93
Hualong Securities Maintains Buy on Yinlun as Liquid Cooling, Humanoid Robot Efforts Pay Off
02:55
Industry 96
Unitree Launches G1+ Humanoid Robot with Six Major Upgrades Across Motion and Perception
02:43
Industry 95
Sealand Securities Initiates Kaidi at Buy on Xinquan Tie-Up for Auto-Robotics Push
01:51
Tech 94
Zinova Bets on Tool-Use Intelligence, Not Dexterous Hands, for Construction Robots
01:43
Tech 93
Boston Dynamics Gives Atlas a Head—Deliberately Not a Human One
Embodied AI FRONTIER - Your Embodied AI Intelligence Hub -