NVIDIA has introduced a clever solution for the testing of autonomous vehicles: a cloud-based environment using photorealistic simulation. This system, dubbed NVIDIA DRIVE Constellation, will offer a safe, scalable, and adaptable way of testing self driving cars.
The system comprises of two distinct servers that together are capable of generating billions of miles of autonomous vehicle testing.
The first server is designed to simulate all the sensors of a self driving car and runs on NVIDIA’s DRIVE Sim Software. It is able to generate photorealistic environments and scenarios ranging from time of day, weather conditions, glare, and different road conditions. This is all aided by its powerful onboard NVIDIA GPUs.
The second server runs NVIDIA DRIVE Pegasus, a complete AI car computer that takes the simulated data from the first server and interprets it as a real autonomous vehicle would. This enables the DRIVE Constellation to generate billions of test miles per year, in contrast with a fleet of 20 real test cars that can only cover 1 million miles per year.
Rob Csongor VP and GM of Automotive at NVIDIA says this technology is able to test custom scenarios and rare corner cases in a fraction of the time and cost that it would take to do so in real world conditions.
Research Analyst Director at IHS Markit, Luca DeAmbroggi, has said that the DRIVE Constellation “will bring us a step closer to the production of self-driving cars.”
DRIVE Constellation will be available sometime in the third quarter of 2018 for early access partners.