In March 2026, AgiBot announced its AgiBot World dataset crossed 1 million real-world robot action trajectories. Every trajectory comes from a physical robot executing a physical task — joint angles, force readings, and camera frames logged at millisecond resolution. This is not simulation data. Understanding this dataset is critical for evaluating Chinese humanoid vendors in 2026. The gap between brands is no longer primarily about hardware. It is about data.
How AgiBot Generates Trajectories at Scale
AgiBot’s Shanghai facility runs 100 humanoid robots simultaneously collecting data. Human operators provide teleoperated demonstrations through a haptic interface. The operator moves; the robot mirrors; every action is logged. One pick-and-place trajectory contains 3,000 to 8,000 time-step records. A complex assembly trajectory — inserting a connector with 2N insertion force — runs to 15,000 records. At 100 robots running 16 hours per day, AgiBot generates 8,000 to 12,000 trajectories daily. Reaching 1 million required approximately 100 to 125 days of continuous operation. The implication: this infrastructure has been running since at least mid-2025.
Why Trajectory Count Matters
Modern manipulation policies train on demonstration data via imitation learning. Policy quality scales with two variables: diversity and quantity. Diversity means the same task demonstrated across different lighting, positions, surface textures, and starting configurations. Quantity drives generalization that diversity alone cannot. With enough demonstrations, the policy network learns physical priors — what contact forces feel like, how objects respond to gravity, when to slow down and apply controlled force. These priors transfer across tasks.
Tesla’s Optimus team targets 1 billion simulation steps. AgiBot’s 1 million real-world trajectories are not directly comparable — real and synthetic data have different noise characteristics — but the scale logic is identical. Data quantity is the primary lever for making robots useful in unstructured environments.
The Compounding Advantage
More data produces better policies. Better policies mean more reliable robots. More reliable robots deploy in more environments. More environments generate more diverse data. Each iteration widens the gap between the data-rich leader and every follower.
AgiBot’s 10,000 cumulative units shipped by March 2026 represent 10,000 data collection endpoints, each contributing trajectories from real customer sites across automotive, electronics, and logistics. Every deployed robot is also a sensor. Unitree has similar shipment volumes but sells to a more diverse base — researchers, developers, hobbyists — who may not feed structured training data back into a central dataset. AgiBot’s enterprise deployment model creates a more controlled data loop.
What AgiBot World Contains
A public subset of 100,000 trajectories was released in late 2025 across 100 manipulation task categories: object sorting, peg-in-hole insertion, cable routing, container opening, drawer manipulation, garment folding, and multi-step assembly. Each category contains 500 to 2,000 demonstrations with 6-axis force/torque data, RGB-D from three viewpoints, and joint state logs. The public release is both a research contribution and a competitive signal — academic labs that build on AgiBot World will publish results that validate the platform.
Implications for Procurement
For industrial buyers evaluating humanoid platforms, the trajectory dataset is a due diligence question. Ask vendors: how many trajectories collected? On what task categories? What fraction from real environments versus simulation? What is your data collection rate per month? A vendor with fewer than 100,000 real-world trajectories in 2026 still requires extensive on-site fine-tuning for each deployment. A vendor with 1 million+ diverse trajectories can deploy policies that generalize out of the box for standard tasks — significantly reducing integration time and cost.
See our Top 40 China Robot Rankings 2025 for the full humanoid competitive landscape and robot manufacturers in China for factory profiles.
Sources
- People’s Daily EN: AgiBot ships 10,000 cumulative humanoid units (March 2026)
- AgiBot World dataset release notes (November 2025)
- ESM China: Humanoid robot shipment data H1-H2 2025
