Predictions about self-driving cars

googlecar

Predictions are hard, especially when they’re about the future. With that said, self driving cars are almost certain. Why? Because going to the hospital is bad. Very bad. Quote: “More than 30,000 Americans are killed each year in road accidents, and another 240,000 or so have injuries severe enough to require hospitalization.” Let’s take a look at how the transition might occur.

Technical and Economic Challenges

If you follow technology, you likely know the major technical challenges have already been solved. Google’s self-driving cars have driven 300,000 miles without an accident. What’s left is figuring out nuances of driver-driver or driver-pedestrian interaction. For example at a four way stop someone waves to let you through. What does the car software decide to do? Drivers informally signal to people with hand waves all the time. Should a car “wave” to a pedestrian with specially designed lights to tell them to cross in front? There’s lots of nitty-gritty human interaction questions to be worked out when you can’t yell or gesture at the driver. But these aren’t blockers.

Predicting the economics is trickier. The radar that sits on top of Google’s self-driving cars costs 75k today, and the overall cost of a car is 150k. But costs are rapidly coming down, for example here. Prices should be in the roughly 50k-100k price range in a decade.

Privacy Challenges

I’ve talked about privacy and self-driving cars before. Self-driving cars will have to record video everywhere they go. Otherwise people will deliberately get into an accident to win a lawsuit. But history shows people give up privacy for convenience every time. It’s a little disturbing to imagine now, but soon we’ll merely shrug knowing every car going down the street is capturing us on video. A related precedent is Google street view, which flared up as a privacy issue at first but was utterly forgotten in no time.

Legal and Insurance Challenges

The biggest hurdles left for self-driving cars are not technical. It’s who gets sued and who pays for insurance. Suppose several pedestrians cross the road without looking and a self-driving car swerves to avoid them and kills a bystander. Who gets sued? Or what if a self-driving car gets into a ten car pileup. Who’s at fault? I think car manufactures, not the human “drivers”, will be sued. The car was driving after all, and car companies have deep pockets.

Lawsuits will force self-driving cars to be far safer than human drivers. But that’s not difficult given 30,000 deaths and 240,000 hospitalizations a year, mentioned above. Driving is one of the most dangerous things people do on a regular basis.

How will this play out? Short term hard to tell. If there’s a really bad crash and some school children die and a self-driving car gets blamed, it could delay things for decades while lawsuits go back and forth. On the other hand, the legal environment right now is highly favorable. California and Nevada are passing laws encouraging self-driving cars. Self-driving cars are cool. Politicians want in. So laws are getting passed. It would be surprising if this favorable dynamic changed. To get adopted new technologies need to be not just better, but way way better. Preventing 500 dead people a week is way way better.

Predictions

To paraphrase science fiction writer Frederic Pohl: predicting the automobile in the 1890 would have been easy, but predicting traffic jams, carpool lanes and smog would have been hard. By analogy predicting self-driving cars in 2013 is easy, but predicting secondary consequences is hard. Yet that’s where the fun is. So let’s give it a try despite the fact that many of the predictions below, like all predictions, will turn out to be wrong. You’ve been warned.

Short term, next 10-15 years:

  • Long haul trucks will be early adopters. Why? Because it’s a multibillion dollar industry which can negotiate insurance and lawsuits. They have a clear economic incentive, as early adopters might have a winner take all situation on the first company to move trucks 24×7 without drivers sleeping. Of course incumbent licensed truck drivers will hate this, so it will require a new entrant into the trucking industry to try it out.
  • Early cars will be expensive and for the rich, minimum 75k. This is a classic technology situation. Cars around 1890-1905 were the same way, until the middle class blockbuster Model T was introduced in 1908, about fifteen years into the automobile era.
  • Taxi companies and drivers will successfully block adoption of self driving cabs short term. A good analogy is Uber, an app which can call a cab, being blocked by entrenched interest lawsuits.

Long term, next 15-50 years:

  • Private car ownership nationwide will drop. Instead we’ll have car services that own and maintain self-driving fleets. You’ll be able to call a car on demand from a phone app. Think Zipcar on steroids. Especially true in cities where calling a car will let you avoid the time sink of parking. The whole purpose of a car is to go where you want, when you want. If you can achieve this without the pain of owning car, paying maintenance, paying insurance and worrying about parking, lots of people will do it.
  • With that said, many people will still drive their own cars. Especially older folks who’ve lived their whole lives driving. Co-existence will be the norm, not replacement. As a side prediction, the nostalgic craze for classic muscle cars will get worse.
  • Commuting to work by driving your own car won’t be commonplace and boring. It will be like a riding a high powered motorcycle is today. Slightly risky. Dangerous with retro charm. Though social pressure will eventually come down hard on human drivers. Traffic accidents are dangerous for innocent bystanders, not just the cool kids. Zero tolerance for alcohol in drivers will become the law.
  • Self-driving carpool lanes will come into being, where cars will go over 100 miles an hour. Car driving algorithms will be combined with GPS traffic control to massively increase traffic flow and speeds on our existing highway infrastructure. Highways will look the same as today, until you pay attention and realize the cars are going way too fast and are too tightly spaced.
  • Parking lots, especially in cities, will be reduced and pushed to peripheral areas. Self-driving cars can drop you off and then self-park. Why waste prime real estate with a parking lot? Sending your car a mile or two away to park will be close enough.
  • Cars redesigned from the ground up for driverless riding will come to market. They won’t even have a steering wheel at all. Being so much safer, the seats could all face to the center (instead of frontwards) so everyone can chat during a drive. Cars interiors will start to look like mobile work cubicles. Chairs will fully recline so you can catch a nap.
  • Many car companies today will go out of business. During a major technology transition, incumbents fail and upstarts take over. Many who survive will become niche players. A good analogy is how Harley Davidson reacted to cheaper asian motorcycle manufactures by becoming a small but successful niche player. It wouldn’t be surprising if retro Mustangs (with steering wheel included!) were the big money maker for a vastly downsized Ford in 2060.

Super long term, 100-150 years from now:

  • Dude, where’s my flying car? Flying cars are a classic sci-fi trope of the future that will never will come to pass, as Calvin and Hobbes complain about below.

calvincar

  • To be clear, personally owned flying cars (really private helicopters) driven by their owners will never ever be mainstream. The whole idea was dumb in retrospect, since it’s obviously a safety nightmare. We’d replace drunk driving teenagers doing donuts in high school parking lots with drunk flying teenagers in personal helicopters zooming above nursery schools. No way. And yet….if we have self-driving and self-flying technology, and we have self-flying fleets safely maintained by corporations that got their start with self-driving cars, we could have self-flying taxi services that really are safe. The trick to making flying cars safe enough to be possible is humans can’t fly or maintain them. So maybe as a rare splurge for the rich, we could see a few flying cars (self-flying personal helicopters). Though I can’t see this ever being mainstream since self-driving cars by that point could drive over 200 mpg on dedicated lanes. Hard to beat the safety and efficiency of computer driven ground transport.

Most of the points above, except the last goofy section on flying cars, are already mainstream predictions. For example see The Economist’s take.  And it’s clear this is a big deal given transportation is the fourth largest sector of the economy, at 11% of GDP. For me, my hope is I can call a self-driving taxi to get around when I’m too old to drive myself. Seems possible.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s