3D LiDAR is lots of very fast data. We're trying to speed up vehicles so we're needing to process it fast. We're sending it through Open Robotics ROS middleware. To do so efficiently requires clever algorithmic work and Eclipse Cyclone DDS ROS middleware with point cloud optimizations.
It requires capturing and processing a million points per second. Each point is a cartesian X, Y, Z as well as reflectivity and intensity. And in parallel we're running deep learning algorithms and model based approaches to enable scene understanding. This is a lot of data and a lot of computation. And this is all so you can enable safer, faster, and more intelligent driving. For example slow the vehicle down when in proximity to people.
We will show videos of vehicles in warehouses using this to travel safely inside facilities at speeds much faster than previously feasible. This HD Perception solution used by warehouse vehicle OEMs is comprised of Box Robotics software with ROS 2 Foxy with Eclipse Cyclone DDS on ADLINK ROScube controllers with Ouster digital LiDAR and AWS RoboMaker services.