Lidar maker Hesai launches Kosmo, a scanner-plus-cloud system that turns real rooms into robot training worlds, and signs Galbot and Dexmal

Hesai Group said in a press release on Oct 14 that it has launched Kosmo, a 3D world data engine that pairs a handheld 8K AI spatial camera, Kosmo One, with a cloud platform, Kosmo World, to turn real spaces into simulation-ready environments for robot training and world-model development.

Kosmo One combines lidar, a one-inch image sensor and an IMU, and Hesai says one operator can scan a typical 100-square-metre room in about 15 minutes; the platform outputs Gaussian-splatting, mesh and depth data, integrates with NVIDIA Isaac Sim and Isaac Lab, and can generate more than 10,000 training trajectories from a single scanned space.

In a Dairy Queen store deployment, building the training environment took 15 minutes instead of about 300 hours of manual modelling and the Sharpa robot reached a 99% task success rate, the company said.

Hesai also announced partnerships with embodied-AI companies Galbot and Dexmal and said cumulative deliveries of its robot actuation modules, which power Sharpa's Wave hand, passed 10,000 by the end of the second quarter, with six-figure shipments expected in 2027.

No price was given. Galbot and Dexmal are both in MEKAWIRE's database, with Dexmal closing Series A and A+ rounds in November 2025, which makes Hesai one of the first lidar suppliers selling data, not just sensors, to China's humanoid makers.