Hesai launches AI spatial camera for physical AI applications

Kosmo integrates lidar, high-resolution imaging, and AI algorithms, including 3D Gaussian Splatting (3DGS) reconstruction technology, to create high-fidelity 3D representations of spaces like campuses and factories. Hesai positions these 3D models as foundational infrastructure for physical AI, serving as scalable training datasets for embodied intelligence and world models. The technology also aims to streamline creative workflows in filmmaking, game development, and cultural heritage by accelerating the creation of production-ready 3D content.

Alongside Kosmo, Hesai showcased its broader Physical AI infrastructure, which includes the JT series mini 3D robotic lidar, the ETX series 6D full-color lidar, and the XT series lidar. The company highlighted that robots are projected to deploy significantly more lidar sensors than automotive applications by 2050, positioning Hesai to capitalize on this trend. The ETX Series lidar, designed for L2-L4 autonomous driving, offers advanced features like chip-level fusion of RGB and time-of-flight data for colorized point clouds, with a detection range of up to 600 meters and 4,320 channels. Hesai's strategy is to build the infrastructure for physical AI across perception, understanding, and execution.

The Kosmo AI spatial camera is slated for official release in the second half of 2026.