# Hesai unveils Kosmo to speed robot training data capture

Digest AI · Robotics & Physical AI · published 2026-10-14T18:27:20Z

Canonical: https://digestai.news/story/hesai-unveils-kosmo-to-speed-robot-training-data-capture

## Summary

Hesai Group launched **Kosmo** on October 14, 2026, a 3D world data engine combining its **Kosmo One** spatial camera with the **Kosmo World** cloud platform. The system converts physical environments into high-fidelity 3D digital spaces for robot training, world model development, and immersive content creation. The **Kosmo One** camera integrates 8K lidar, a one-inch imaging sensor, and inertial measurement units, achieving millimeter-level geometric accuracy. Hesai claims it can scan a 100-square-meter indoor space in about 15 minutes with a single operator, processing data up to five times faster than comparable tools, though results vary by scene complexity.

Hesai demonstrated Kosmo’s efficiency in a **Dairy Queen** deployment, reducing manual modeling time from 300 hours to 15 minutes. A robot’s task success rate improved to 99%, with failures dropping to one-tenth of previous levels. The company also highlighted cost savings in filmmaking—scanning and reconstructing a 10,000-square-meter European streetscape takes one week with Kosmo, compared to two months for traditional set-building. For world models, a single scan yields over 10,000 training trajectories, offering 100x more data at lower cost than manual collection. Strategic partnerships include **Galbot**, **Dexmal**, **Pixar’s RenderMan**, and **Amap**, with applications spanning robotics, film, and location services.

## Key points

- Kosmo combines **Kosmo One** camera and **Kosmo World** cloud to digitize real-world spaces for robot training
- Scanning a 100-square-meter space takes 15 minutes, with 5x faster processing than competitors, per Hesai
- Dairy Queen deployment cut manual modeling time from 300 hours to 15 minutes, improving robot success rate to 99%

## Why it matters

Kosmo addresses a bottleneck in Physical AI: affordable, high-fidelity real-world data for training. Faster capture and lower costs could accelerate robotics deployment and reduce film production timelines.

## Sources

1. [Hesai Launches Kosmo to Turn Real-World Scans Into Robot Training Data](https://unite.ai/hesai-launches-kosmo-to-turn-real-world-scans-into-robot-training-data) (Unite.AI, 2026-10-14)

## Cite

Digest AI, "Hesai unveils Kosmo to speed robot training data capture", 14 October 2026, https://digestai.news/story/hesai-unveils-kosmo-to-speed-robot-training-data-capture

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/hesai-unveils-kosmo-to-speed-robot-training-data-capture.json