The first big commercial deployment of driverless car technology is coming not in the streets of Silicon Valley but in the arid and sparsely populated Pilbara region of Australia. That's where the large mining conglomerate Rio Tinto has rolled out fleets of all-driverless trucks at two iron ore mines, according to a report by Jamie Smyth at the Financial Times.
Rio Tinto tells Smyth that the driverless transition has improved performance by 12 percent, mainly by "eliminating required breaks, absenteeism and shift changes."
GPS guides the trucks and allows them to deliver iron ore 24/7, 365 days a year, without the kinds of breaks and handover periods that human drivers would need. The GPS navigation system is backstopped by a team of human operators working remotely from Perth, hundreds of miles away. Not only does this reduce the total number of humans who are needed to run the trucking operation, but it eliminates the need to employ those humans in the remote and desolate mining country. A mine needs to be located where the ore is, and you often end up needing to pay a premium to recruit workers to ore-adjacent locations. Remote workers, by contrast, can live in a nice suburb of a midsize city.
Why driverless is coming to mines first
But of course mining operations are hardly the only companies that could save money by replacing human drivers with robots. The big difference is that the mine setting offers a huge advantage that most companies can't match — there are no human drivers on the road.
As David Roberts has written, one of the biggest technical challenges in making a general-purpose autonomous vehicle is that it would have to deal with all those crazy human beings. If all the cars were autonomous and networked, they would interact and communicate in predictable ways. But for that to happen in a normal transportation context, you would first need a transition phase in which autonomous cars co-exist with human-piloted ones long enough for them to gain trust and traction.
Rio Tinto doesn't have that problem. It controls the entire site, and can make the transition to an all-autonomous fleet all at once. There's no "transition period"; there's just a transition.
Commercial settings will go driverless first
To most people, the current face of driverless car technology is a pretty normal-looking Google car cruising the streets of the Bay Area. But most of the concrete progress over the next few years is likely to happen in areas that are more conceptually and geographically distant from the core functions of the modern automobile.
Rio Tinto is only doing this at two mines so far, and it has a lot more mines. There are also other mining companies, and other companies in fields like lumber harvesting that involve a similar need to haul lots of heavy stuff around privately owned land. You probably won't hear them called "driverless cars," but warehouse robots that move things around a confined space while interacting with a limited set of other robots are also starting to come into use. Automation of light trucks and specialized vehicles on farms and ranches would be an interesting next step. These are in some ways closer to the generic "driverless car" case, but they also involve private land and not many people.
After that, who knows — zoos and amusement parks? Airports? Transit buses with dedicated lanes? The point is that rather than seeing a slow trickle of autonomous cars start to appear on busy streets (the way we have seen with electric cars, for example), you are likely to see mass adoption of driverless vehicles in settings that are somewhat marginal to the core market for light vehicles. Then if the technology succeeds, it will gobble up more and more slices, along the way establishing the kind of clear track record and public understanding that would be necessary to move into the core area.