A number of self-driving cars sporting NVIDIA’s autonomous driving technologies were present at GPU Technology Conference in San Jose this week. However, none was available for a test drive — a sensible move by NVIDIA in the light of the fatal Uber self-driving car accident the week before.
But NVIDIA did reveal the NVIDIA Drive Constellation, a cloud-based for testing autonomous vehicles using photorealistic simulation. This will create a safer, more scalable method for bringing self-driving cars to the roads.
Available to early access partners in Q3, the platform is based on two servers — one that runs the NVIDIA Drive Sim software to simulate a self-driving vehicle’s sensors, and a second that contains the NVIDIA Drive Pegasus AI car computer.
The simulation server is powered by NVIDIA GPUs, each generating a stream of simulated sensor data, which feed into the DRIVE Pegasus for processing.
Driving commands from Drive Pegasus are fed back to the simulator, completing the digital feedback loop. This “hardware-in-the-loop” cycle, which occurs 30 times a second, is used to validate that algorithms and software running on Pegasus are operating the simulated vehicle correctly.
Drive Sim software generates photoreal data streams to create a vast range of different testing environments. It can simulate different weather such as rainstorms and snowstorms, glare at different times of the day and night, and different types of road surfaces. Dangerous situations can be scripted in simulation to test the autonomous car’s ability to react, without ever putting anyone in harm’s way.
“We use extreme caution, and the best practices that we know in testing our cars,” he said. “First of all, it’s of course a safety concern, because our engineers are actually in the car. So it’s something we take incredibly seriously,” said Jensen Huang, CEO of NVIDIA.