Hesai Group has launched Kosmo, a 3D spatial data platform that converts real-world environments into detailed digital models for training artificial intelligence systems and developing intelligent robots. The Shanghai-based lidar maker announced the platform this week as a tool for building the synthetic environments that robot-learning pipelines require.

Kosmo combines three measurement functions in one platform: lidar ranging, high-resolution imaging, and 3D reconstruction. Together they capture a physical space and turn it into a structured digital environment — the format that AI training pipelines consume when a robot must learn to operate in that space.

What problem does Kosmo address?

Robot developers face a persistent data bottleneck. An AI-driven robot needs large volumes of environment data before it can act in a factory, warehouse or inspection route, and that data must represent real geometry rather than idealized simulation. Hesai positions Kosmo as the capture step: it digitizes actual environments so the resulting models ground robot training in measured spatial structure.

The platform pairs lidar depth data with high-resolution camera imagery and fuses both through 3D reconstruction algorithms. The output is a digital environment suitable for training AI systems and for developing intelligent robots, according to the company.

Hesai did not publish specifications for the Kosmo platform — no range figures, resolution numbers, accuracy classes or throughput rates appear in the launch announcement. Buyers evaluating the platform against alternatives will need to request measured performance data and the test conditions under which it was obtained.

The absence of published specs marks this as an early platform launch rather than a shipping instrument datasheet. Until Hesai releases concrete figures, the measured-versus-claimed distinction that drives metrology purchasing decisions remains open.

Where would it be deployed?

Hesai names two application areas for Kosmo: automated inspection and smart manufacturing. Both demand faithful spatial models of the operating environment.

In automated inspection, a robot trained on a Kosmo-generated model could navigate and assess equipment whose geometry the platform previously captured. In smart manufacturing, digital environments from Kosmo could support robots learning layout-specific tasks — pick, place and move operations tied to a specific line configuration.

The application logic follows directly from the physics of the sensor mix. Lidar supplies geometric depth independent of ambient light; high-resolution imaging supplies texture and feature detail; 3D reconstruction binds both into a model an AI system can train against. Hesai's contribution is the integration of that chain into a single spatial-data platform.

Who is Hesai?

Hesai Group is a China-based lidar manufacturer with an established position in automotive and robotics sensing. Kosmo extends the company from component supply — lidar units — into a data-platform layer that serves downstream AI and robotics development. The move ties Hesai's ranging hardware to the training-data workflow that increasingly determines robot deployment timelines.

The takeaway: Hesai has packaged lidar, imaging and reconstruction into one platform for turning real environments into robot training data, but the launch ships without published specifications.

The open question for adopters is compliance-grade performance: whether Hesai will publish the accuracy, resolution and throughput figures — with stated test conditions — that inspection and manufacturing users need before committing a production line to models built on Kosmo data.