From the department of “I read the New Yorker so you don’t have to,” here’s the crucial few sentences of a long article:
Left to its own devices, Thrun says, [the driverless car] could go only about fifty thousand miles on freeways without a major mistake. Google calls this the dog-food stage: not quite fit for human consumption. “The risk is too high,” Thrun says. “You would never accept it.” The car has trouble in the rain, for instance, when its lasers bounce off shiny surfaces. (The first drops call forth a small icon of a cloud onscreen and a voice warning that auto-drive will soon disengage.) It can’t tell wet concrete from dry or fresh asphalt from firm. It can’t hear a traffic cop’s whistle or follow hand signals
No, I don’t actually want to be too hard on Burkhard Bilger’s article, which is a nice overview for those who are just starting to hear about driverless vehicles. But I assume that the few, the proud, the readers of Sandor at the Zoo know the widely reported optimistic bits.
I don’t have a lot of commentary to add. Google’s cars have driven a long way without an accident, but they’ve done so on dry roads et cetera. That’s more or less what I thought, but it was nice to actually hear confirmation. The big question right now is “are we at the stage of spending a few years ironing out the last details or are we just starting to tackle the really hard stuff?” I don’t know the answer. It may even be that Google doesn’t really know the answer.
I do caution that automatic cars can’t be a half-way solution. As soon as you start telling people that 90%+ of the time, they don’t have to touch a steering wheel, you lose any ability to practically use humans as a fail-safe. The drivers will be reading, or dozing, or drunk, or will just be so rusty at driving — having not done any in months or years — that they can’t handle a challenging driving situation. You can’t fix this problem legally, not without undoing most of the things that make driverless cars really transformative.
So the cars need to get to the point where they handle those challenging driving circumstances much better. Then they need to overcome legal challenges. Then they need to be cost-effective. And then all of our lives will be better. Driverless cars are clearly coming, even if we have a huge step back and have to totally rethink Google’s LIDAR sensors because they’re simply incompatible with rain, for example. But 5 to 10 years out, which are the numbers typically being thrown around, is still a very speculative estimate.