Hesai launches Kosmo, a 3D mapping camera that feeds spatial data to humanoid robots

The wine-bottle-sized device combines lidar and cameras to collect real-world spatial data, targeting the training-data gap for AI-powered robots.

Hesai Group unveiled Kosmo on October 14 at a launch event at its Shanghai factory, a wine-bottle-sized device that pairs lidar sensors with cameras to help developers collect real-world spatial data more efficiently. The target market is clear: the AI systems being trained to power humanoid robots, which need physical spatial data the way language models need text.

The company sees a wide gap between today's robots and the productive machines the industry is chasing. Co-founder and CEO David Li told attendees that a significant shortfall in high-quality spatial data keeps AI-powered robots from understanding and interacting with the real world, leaving most of them confined to low-impact functions such as greeting guests, waving, or dancing. Robots remain limited in their ability to solve real productivity challenges, Li said.

Li laid out a bigger ambition: making Hesai the underlying technology platform for the robotics industry, a tier-one supplier in the mold of Bosch in the automotive world, or perhaps even beyond. He declined to give revenue projections for Kosmo but said Hesai has already signed agreements with Chinese robotics companies to use the device.

Hesai is also moving deeper into the robot body itself. Li said the company is developing actuators, the components that convert electricity into motion and account for more than half of a robot's material costs, with demand growing for versions that are lighter, smaller, and more precise. The actuators are already in use by Sharpa Robotics, a startup founded by Hesai's own co-founders, inside its flagship Sharpa Wave robotic hand. A Sharpa Wave robot has been autonomously serving ice cream at a Dairy Queen outlet in Shanghai, and Hesai says Kosmo's detailed mapping helped it reach a 99 percent task success rate.