Artificial intelligence news.
Straight from the original announcements.

Hesai Launches Kosmo, a 3D World Data Engine for Physical AI Training

October 14, 2026, 20:04By the AI News Bites Editorial TeamSource: PR Newswire
Hesai Launches Kosmo, a 3D World Data Engine for Physical AI Training
Image: PR Newswire

Hesai Group has launched Kosmo, a 3D world data engine that combines an AI spatial camera with a cloud platform to turn real environments into training data for robots and world models. According to the company, the system targets a key bottleneck in Physical AI: access to high-quality real-world data at scale.

Key points
  • Hesai launched Kosmo, combining the Kosmo One 8K AI spatial camera with the Kosmo World cloud platform.
  • According to Hesai, one operator can scan a typical 100-square-meter indoor space in about 15 minutes.
  • In a Dairy Queen deployment, Hesai says the Sharpa robot's task success rate reached 99%.
  • Hesai announced partnerships with Galbot and Dexmal and showcased work with Pixar's RenderMan and Amap.
  • Kosmo integrates with NVIDIA Isaac Sim, Isaac Lab, Unreal Engine, RenderMan and Houdini.

Camera and cloud platform

Kosmo has two parts. Kosmo One is described by Hesai as the world's first 8K ultra-high-definition AI spatial camera. It integrates high-resolution lidar, a one-inch imaging sensor and a high-precision inertial measurement unit (IMU) to capture geometry, color, texture and camera position with microsecond-level synchronization.

Reconstruction uses 3D Gaussian Splatting (3DGS), which according to Hesai delivers millimeter-level geometric accuracy. The company says a single operator can scan a typical 100-square-meter indoor space in approximately 15 minutes. The second part, the Kosmo World cloud platform, processes captured data into 3D models at speeds of up to approximately five times those of comparable products, according to Hesai.

Kosmo World generates 3DGS, mesh and depth data in a single reconstruction. Generative AI fills unscanned areas, and AI-powered semantic segmentation identifies individual objects. Views can be adapted to a robot's camera characteristics, which Hesai says helps narrow the gap between simulation and real-world operation.

Robot training and world models

Hesai points to a deployment at a Dairy Queen ice cream store. Creating a 3D training environment there previously required approximately 300 hours of manual modeling by a professional team, the company says. With Kosmo, scanning and reconstruction took 15 minutes. In this deployment, the Sharpa robot's task success rate reached 99%, and the failure rate fell to one-tenth of its previous level, enabling the robot to begin operating in an actual store.

For world model development, Kosmo provides geometrically accurate 3D scenes with precise camera poses, depth information and consistent multi-angle views. From a single captured space, Hesai says Kosmo can automatically generate more than 10,000 training trajectories, giving developers more than 100 times as much high-quality training data for the same cost.

At the launch event, Hesai also announced partnerships with embodied AI companies Galbot and Dexmal.

Film, heritage and maps

Hesai also positions Kosmo for creative and commercial uses. In film, a location can be captured once and reused from previsualization through postproduction, including in tools such as Unreal Engine. Hesai says a 10,000-square-meter European streetscape that would take two months to build as a physical set can be scanned, reconstructed and edited in one week, with potential cost savings of up to 80%. Workflows built around Pixar's RenderMan are already in use, and RenderMan has released an official tutorial. Hesai is also working with Yuemo Studio on AI-assisted filmmaking.

For cultural heritage, Hesai is working with TRIC, a United Nations partner, on an immersive digital artwork of Du Fu Thatched Cottage in Chengdu. With Amap, reconstructions of the Gravity Art Museum and PIFO Gallery are live in the "Flying Street View" feature. Hesai says Amap reaches nearly one billion monthly active users, and the stated aim is to extend navigation from finding a store to locating a product inside it.

Open ecosystem and business models

Hesai describes Kosmo as an open spatial intelligence ecosystem, with open access to raw capture data and 3D assets and integration with NVIDIA Isaac Sim, Isaac Lab, Unreal Engine, RenderMan and Houdini. It supports two business models: integrated hardware and software solutions for customers in robotics, media and other industries, and services that produce real-world spatial data at scale for world model and Physical AI developers.

The launch is part of a wider push. Hesai says its lidar deliveries have doubled for nine consecutive years and that it holds a leading global position in passenger-vehicle ADAS and robotaxis. It is also expanding robotic actuation: its modules power Sharpa's Wave dexterous hand, cumulative deliveries exceeded 10,000 units by the end of the second quarter of 2026, and shipments are expected to reach six figures in 2027.

What it means

If Hesai's figures hold in wider use, Kosmo could cut the time and cost of building simulation environments for robot developers, and give studios and map platforms a faster way to digitize real places. The performance claims come from the company and rest on its own testing and product comparisons. CEO Dr. Yifan "David" Li calls 3D world data and robotic motion Hesai's "second growth driver."

Hesai GroupKosmoKosmo OneKosmo WorldGalbotDexmalroboticsphysical AIenterprise AIpartnerships

Related news

FF EAI Robotics Ecosystem Adds NYU's Carolyn Kissane to Its Board
BusinessNovember 20, 2026, 03:59

FF EAI Robotics Ecosystem Adds NYU's Carolyn Kissane to Its Board

FF EAI Robotics Ecosystem Inc. (NASDAQ: FFR) has appointed Carolyn Kissane, Associate Dean of New York University's Center for Global Affairs, as an Independent…

Business
BusinessNovember 17, 2026, 08:20

Boyuan Capital closes RMB 1.1 billion Fund II for industrial AI

Boyuan Capital has completed the final close of a Fund II worth approximately RMB 1.1 billion, with a focus on industrial AI, embodied intelligence and the global…

Galileo Deploys C1-W Robot Dog for Warehouse Patrols in Poland
Products and toolsNovember 16, 2026, 09:10

Galileo Deploys C1-W Robot Dog for Warehouse Patrols in Poland

Galileo Robotics has deployed its C1-W quadruped robot for automated warehouse inspection and security patrols at a large logistics park in Poznan, Poland. The company…