Audi’s engineers highlighted during a conference last week in California, held by chipmaker Nvidia Corp., an unforeseen problem with their “Jack” driverless car prototype.
Back in January, during a trip that took the A7 sedan on a 550-mile trip to the International CES technology expo in Las Vegas, Jack encountered an interesting obstacle on a remote stretch of desert highway – a piece of tumbleweed stuck to the front of the car. While that would be an event that has no meaning for us, for “Jack” – which uses numerous, lasers, sensors and cameras to guide itself in traffic, the event was a potentially crucially hazard. Staying stuck to the grille for 10 or 15 miles it blocked some of the autonomous car’s sensors, according to Daniel Lipinski, a senior engineer at Audi. Nothing happened and the car went on, but the occurrence shows one of the biggest autonomous driving challenges – how can a self-driving car react to the unexpected. “We can’t program a car to do every behavior,” commented Jen-Hsun Huang, CEO of chipmaker Nvidia Corp.
The problem has been on the minds of Nvidia and other tech companies, as well as automakers for years, with the solution being the field of “deep learning,” a sci-fi like strategy that aims to teach computers to process data in a way resembling the human mind. Nvidia seeks to develop computers that might “learn the behavior of driving over time and can be updated over time to be smarter and smarter and better and better at driving,” according to Huang.
Via Automotive News