Note To The Skeptics: The World Will Change For Self-Driving Cars – Forbes

Posted: Sunday, October 26, 2014

Every new technology has skeptics at first, and self-driving cars are no exception. I am fully willing to hear these skeptics out, but the most common complaints today are just far too short-sighted to be taken seriously. What they miss is that big new technologies don’t just adapt to the world, but the world adapts to them.

A new piece by Lee Gomes at Slate provides an excellent example of how the skeptics go wrong. He rehashes some well-worn complaints which I’ll address one-by-one, and show how they reflect a far too narrow and static vision of the future.

First, Gomes complains that self-driving cars will require extremely detailed maps that are expensive and difficult to create:

Since maps are one of the engineering foundations of the Google Google car, before the company’s vision for ubiquitous self-driving cars can be realized, all 4 million miles of U.S. public roads will be need to be mapped, plus driveways, off-road trails, and everywhere else you’d ever want to take the car.

This assumes that the information will continue to be pulled by the car companies rather than being pushed by the government. If this technology can demonstrate that it works and saves lives, you can imagine detailed up-to-date mapping of the roads being a task the government takes. After all, you can’t build or alter a road without government approval already, and they spend billions of dollars of years keeping them maintained. So in terms of the scope of the work the government already does on roads and automobile infrastructure, this would be only a marginal increase in cost and responsibility for them.

This is closely related to Gomes’ next complaint, which is that the mapped information needs to be highly up-to-date:

“Considering all the traffic signals, stop signs, lane markings, and crosswalks that get added or removed every day throughout the country, keeping a gigantic database of maps current is vastly difficult. “

Again, you don’t put up a traffic signal without government approval, so having the government make a widely available digital record of this before the change hits the real world doesn’t seem so drastic or unimaginable. Gomes argues signs and signals are change far too quickly, and quotes an electronic road sign company that says their portable construction signs are often “simply towed to a site and turned on”. Is it really so difficult to imagine these sign companies being required to upload the GPS coordinates of these signs to a government database before they turn them on?

All these mapping concerns make a huge assumption about the world, which is that roads and signals will remain designed to be seen by humans. Instead, roadways and signage could be required to emit some kind of signal that self-driving cars can easily detect, some kind of digital lighthouse… Or something like that. I’m not an engineer, so I won’t get bogged down in the details of this. The point is it’s crazy to think that the signs and signals that inform drivers must forever be designed with their visibility to humans and not machines as the only concern. It’s easy to imagine how even roads could be adapted to be more easily seen by machines. Is there a kind of paint that could be used on the lines that machines can see much more easily? Again, just spitballing here. But imagining ways the world could adapt to self-driving cars is a better way to think about current difficulties than assuming they will remain insurmountable or only solved by better machines.

The author also complains that self-driving cars can’t operate in the snow. This is a humorously ironic complaint since the inability to drive in the snow was one of the reasons why automobile skeptics claimed that they would never replace horses. In the late 1800s and early 1900s there were a lot of environmental problems that gave horses an advantage. Early roads were made of dirt and would turn to mud that cars could not handle but horses could. It’s clear in retrospect that a massive change to infrastructure was necessary for cars fully replace horses. Evenly graded stone or paved roads would need to become ubiquitous for cars to truly replace horses. When you consider the scale of the change that society needed to make the move from horses to automobiles it really makes the changes self-driving cars require seem quite minor in comparison.

Automobiles weren’t the only technology that required big changes in infrastructure to really replace existing technologies. Compare the work that needed to be done for electricity and telephones to replace the oil lights and telegrams. In comparison self-driving cars mostly need roads, signs, and other cars to simply communicate with them better. Not that big of an ask when you consider the tens of thousands of lives that will saved.

Complaining that self-driving cars can’t see the existing infrastructure as easily as humans is like complaining that automobiles can’t handle dirt roads as easily as horses. It’s a backwards looking complaint that wrongly assumes new technology must adapt to a static world instead of a dynamic world adjusting to the technology.

A robotic Volkswagen Passat shown at Stanford ...

A robotic Volkswagen Passat shown at Stanford University is a driverless car (Photo credit: Wikipedia)




Write a Reply or Comment:

Your email address will not be published.*