InSpatio's system integrates automated physics checks, policy rollouts, and motion planning to produce closed-loop trajectories across hundreds of parallel environments.
The company InSpatio claims that robots currently lack both the volume and the variety of data required for advanced learning. To overcome these limitations, the system provides 100,000 hours of training a day, totaling 1,000,000 hours a month.
InSpatio defines three delivery levels for its technology: L1 for navigation, L2 for dexterous grasping with dense occlusion and millimeter-level geometry, and L3 for contact and articulated dynamics like opening drawers. The company states that the advantages of simulation grow as these delivery levels increase.
Despite the use of simulations, InSpatio notes that final policies are still checked on real robots. The China Academy of Information and Communications Technology reported that more than 70 robot training grounds existed as of June 2026. That same academy announced plans for more than 40 additional robot training grounds.
Hesai's Kosmo 3D world data engine can scan a 100-square-meter room in approximately 15 minutes. This engine produces a 3DGS reconstruction at a millimeter level. These developments are being tracked as of 29 October.
