BERLIN: German car parts manufacturer Continental is now operating its own supercomputer with the goal of accelerating the development of driver assistance systems as well as robot vehicle software.
Continental's fleet currently travels about 15,000km per day - creating roughly 100 terabytes of data. One of the supercomputer's tasks is thus to crunch those numbers and evaluate what they mean.
The supercomputer based in Frankfurt is run with technology from US company Nvidia. Thanks to the supercomputer, systems based on artificial intelligence can be taught faster with large data amounts.
The information also helps train the AI system's neural networks.
It should, for example, help ensure that a vehicle can master a lane change or a roundabout as smoothly as a human would. "Nvidia's high-performance technologies have made it possible for us to shorten processes that used to take weeks to hours," said Christian Schumacher, responsible for driver assistance systems at Continental.
The simulation then comes into play primarily when testing the solutions. "(A distance of) 15,000km a day sounds a lot, but it doesn't cover by far what you need to be able to say that you can drive safely with such systems in every corner of the world," said Schumacher.
"In the real world, many things happen that you can't necessarily foresee." For example, you could drive down a road in sunshine and then simulate rain or snow for the sensors. This accelerates development and saves costs, he said.
The development of assistance systems and technology for autonomous driving involves very large amounts of data - "and our expectation is that this will increase," said Nvidia manager Philippe van den Berge.
