Innoviz will deliver its InnovizTwo LiDAR and perception software for a highway operational design domain, running on an NVIDIA-powered compute platform.
Innoviz Technologies has announced the development of a LiDAR-based perception stack with a top 10 global automotive OEM. Under the collaboration, Innoviz will develop and validate a complete LiDAR-based perception stack for a highway operational design domain (ODD) to explore potential for the OEM's future programs.
The project includes Innoviz's InnovizTwo LiDAR solution and perception capabilities powered by NVIDIA hardware. The perception stack Innoviz will build for the highway ODD represents real Physical AI core functions, including:
- Point Cloud Plus (PC+) and Obstacle Detection: classifies each pixel in the scene as drivable or non-drivable obstacle, flagging in-path hazards such as tires, lost cargo, and Euro pallets along the way.
- Object Detection: identifies objects and road users in the scene, including pedestrians and vehicles.
- Object Classification: sorts detected objects into categories including pedestrians, cars, trucks, and two-wheelers.
"Developing a complete LiDAR-based perception stack for highway driving is the clearest signal yet of where this industry is headed," said Omer Keilaf, CEO and Founder of Innoviz . "We built Innoviz to co-design hardware and software from day one, because that's what we believe production autonomy demands. Delivering that stack to automotive-grade standards is Physical AI doing its job, offering a precise, real-time 3D understanding of the world."
See also:
INTERVIEW: Hardware vs software: The road to autonomous driving
