Hesai doesn’t just want to sell lidar

Image source Hesai Technology Text | Xiao Man Editor | Yuan Silai "We did not build another Kosmo to sell lidar." Before the Kosmo press conference, Li Yifan made this clear. Kosmo is a three-dimensional world data engine recently released by Hesai Technology. It consists of the world's first 8K ultra-high-definition AI space camera Kosmo One and the AI-driven cloud platform Kosmo World. The former is responsible for collecting real-world spatial information, and the latter reconstructs this information into a three-dimensional scene that can be understood, interacted and trained by AI. This is an extension of Hesai beyond its main business of lidar. In the past, Hesai mainly provided perception hardware to cars and robots; now, it hopes to further participate in the training process of physical AI and transform the real world into data assets that can be used by AI. This is also Li Yifan’s first step from a component provider to a core infrastructure provider for physical AI. From lidar to physical AI core infrastructure For physical AI to truly move toward scale, at least two problems need to be solved: how robots complete actions in the real world, and how AI obtains enough and good enough real-world data. Hesai is arranging two businesses around these two issues: robot motion engine and three-dimensional world data engine. Kosmo is the first systematic product implementation of the latter. Image source: Hesai Technology Kosmo One is the data portal of this system. It integrates high spatial resolution lidar, 8K ultra-high-definition imaging system, one-inch outsole CMOS and high-precision IMU into one device to simultaneously collect spatial geometry, color, texture and device posture information of the real environment. It attempts to lower the threshold for spatial acquisition. Kosmo One supports functions such as instant scanning, smart guidance, and one-click resume scanning. In a typical indoor scenario, it takes about 15 minutes to complete a scan of 100 square meters of space. However, scanning a realistic three-dimensional scene is only the first step. For embodied intelligence, scenes require more than just seeing