At its recent AI Day event in Shanghai, Chinese tech firm Hesai introduced Kosmo, a sophisticated platform that transforms physical locations into high-fidelity 3D digital assets. The system aims to provide critical real-world data infrastructure to advance physical AI capabilities.
- Kosmo converts real-world spaces into labeled 3D digital models using lidar and AI.
- Open integration with leading simulation and CGI platforms supports broad ecosystem use.
- Aims to fill data shortages slowing development of adaptive physical AI systems.
What happened
On October 14, 2026, Hesai announced the launch of Kosmo during its AI Day event in Shanghai. Kosmo is a 3D world data engine that captures physical spaces using lidar, image, and motion sensors to create highly detailed, AI-enhanced digital replicas. These digital assets are designed for use in robot training, world modeling, and creating immersive digital content.
Hesai demonstrated Kosmo's capabilities through applications spanning robotics, film and media production, cultural tourism, and spatial commerce, highlighting the platform's versatility. The company also emphasized Kosmo’s open-access model that allows users to obtain raw spatial data and processed 3D assets integrated with major platforms like NVIDIA Isaac Sim, Unreal Engine, and Houdini.
Why it matters
Physical AI is evolving beyond simple programmed responses toward robots and systems that can perceive, reason about, and dynamically interact with the physical world. However, unlike digital AI, which can leverage massive online datasets, physical AI is hindered by the lack of accessible, accurate real-world spatial data for training and development.
Kosmo tackles this critical bottleneck by providing scalable, affordable access to high-fidelity 3D spatial environments enriched with object labeling and AI-generated scene completion. This significantly enhances the ability to develop embodied AI that can operate effectively in real-world conditions, accelerating innovation in robotics and other spatially-aware technologies.
What to watch next
Following its launch, the success of Kosmo will depend on its adoption across robotics, media, cultural, and commercial ecosystems. Hesai’s collaborations with partners like Galbot and Dexmal indicate a focus on pushing embodied AI advancement, potentially positioning Kosmo as a foundational technology for physical AI development in China and beyond.
Monitoring how Kosmo’s integration with simulation tools and real-world applications matures will be crucial. The broader industry’s embrace of open access to spatial data could drive new innovations in robot autonomy, immersive experiences, and location-based digital services, addressing a long-standing data gap in physical AI research.