Highway congestion and bumper to bumper cars for commuters
This week at CES 2019 in Las Vegas NVIDIA revealed its collaboration with HD mapping companies to develop a system that allows autonomous vehicles to localise themselves on a road map.
NVIDIA introduced Drive Localization this week, a system that can be used by an autonomous vehicle to "pinpoint its location within centimeters so it can understand its surroundings and establish a sense of the road and lane structures". When road markings are not clear, this system helps cars understand what exactly is happening on the road so that they can respond and react accordingly.
Drive Localization is powered by the NVIDIA Drive Xavier System-on-chip, a processor specially designed for autonomous vehicle driving that can perform 30 trillion operations per second. Instead of relying on lidar technology, Xavier uses various sensors on the vehicle (cameras, satellites, speedometers, for example) to collect the information needed to determine a vehicle's precise location and orientation.
This process supports NVIDIA's goal of creating a system that reaches level 2+ autonomy, a development which NVIDIA calls Drive AutoPilot. Level 2 autonomy means that a vehicle system is capable of steering, accelerating, and deceleration while a human driver monitors the vehicle's movement for safety. NVIDIA has allegedly accomplished this and a little extra.
The Xavier processor, in addition to determining vehicle location up to the centimeter, is also the brain behind merging, lane changing, and even driver monitoring. Tesla's Autopilot has several of the same capabilities, placing it also between levels 2 and 3 of autonomous driving.
The system is expected to be employed in commercial vehicles beginning in 2020.
