As for real hurdles waiting for self-driving cars, EE Times chief international correspondent Junko Yoshida has come across several astute observations from industry observers.
From start-ups to leading tech companies, opportunities abound for everyone to pile onto the self-driving bandwagon. But there are endless questions about driverless cars.
How are we going to test AI in driverless cars? Do we even know what to test?
Luca De Ambroggi, principal analyst, automotive semiconductors at IHS, told us, “Just as we certify a human driver [with a driving test], the [automotive] industry needs a set of standards or procedures they can use to certify AI–for safety.”
How do we implement autonomous car SoCs, and how do we know the implementation is correct and safe?
Kurt Shuler, vice president of marketing at Arteris, observed, “Design teams today are combining industry-standard CPU clusters (like ARMs) with their own custom hardware accelerators into the SoCs that are the brains of these embedded devices.” He pointed out: “The first challenge for design teams is how best to implement their algorithms and ‘partition’ their implementation in hardware or software.”
Equally challenging in Shuler’s mind is how to qualify these devices for automotive use and adherence to the ISO 26262. Remember, design teams are in essence developing “super computer SoCs with highly specialised IP accelerators,” he noted. He believes some adaptation of ISO 26262 will be needed “to determine diagnostic coverage methods and goals for multicore devices, including those that use cache coherency.”
Automakers’ eagerness to get a piece of the action of the ride-sharing business is accelerating their timetable for driverless cars. Really?
At least that’s what Mobileye seems to believe. They explain that automakers are eagerly buying into the concept of “shared” autonomous vehicles pursued by ride-hailing companies like Uber, as they see it a ‘test bed’ for autonomous cars.
The theory, however, is based on a tenuous premise. Why would American consumers–who essentially prefer driving their own cars to public transportations–suddenly turn eager for robotic cars that will probably only run certain, pre-determined routes?
This is a great social experiment. It might be worth trying. Could it be a real boon to the business of the automotive industry? I’m not so sure. It could remind a lot of people of the urban trolley systems that quietly died out a half-century ago.
Besides, for those of us who don't want to drive, if we already have a ride-hailing service, why would we need a driverless car for?
The best comment this week came to my way from Robert Hollingsworth, a former semiconductor company executive who is now an angel investor living in Austin, Texas. He forwarded me a copy of an Esquire magazine piece entitled “Truth about the Future of Cars” with a note saying:
“I thought that you might appreciate it. It reinforces my conviction that driverless car is going the same route as the paperless office and perhaps 3D TV. Just because it seems possible doesn't mean it should be done.”