In my previous article, I listed the companies that are currently working on self-driving tech, how close are they to market, and what is their approach. Now, I’ll try to make a few educated guesses about what the future holds. I’ll reason my conclusions, but I’m a regular guy with no insider information and no means to predict the future, so feel free to call my bulls**t in the comments.
Proposed timeline of events
- Present: Several car brands have some form of “Level 2” autonomy, which means that the cars can autonomously cruise on highways but the driver must be prepared to regain control if something unexpected happens.
- End of 2017: Tesla will make their hyped coast-to-coast autonomous drive. The human driver will never touch the wheel.
- 2018: The “cruise control” features of several car brands will slowly transition into “Level 3” autonomous driving. That means that the user will receive signals to regain control well in advance, but it will still be forbidden for drivers to multitask (text, watch a movie, read a book etc) while “driving”.
- 2019: Waymo, Tesla and other automakers will have their “Level 4” autonomous cars ready for mass-market. They will retail for about $8,000 more than their “manual” counterparts. Regulations will still require a human, capable driver to be behind the wheel.
- 2020: Regulations in several US states and some European states will start to catch-up. They will still require than a sober adult is behind the wheel, but there won’t be punishing for multitasking while driving in autonomous mode. The cars will need a way to provide that information to the police.
- 2021: Fully autonomous, “Level 5” cars arrive, as an evolution to the “Level 4” cars, although there is no real difference because of regulations.
- 2022: Every US state and every developed country has regulations pertaining “semi-autonomous driving” that allows multitasking while driving in autonomous mode. Every car manufacturer has a high-end model with “Level 4” option.
- 2023: California regulates fully autonomous cars. While in autonomous mode, they can be used by anyone, even if they don’t have licence or they are in the backseat.
- 2023: Uber, Lyft and other private companies buy fleets of fully autonomous cars, leaving many Californian cab drivers without a job. Other states and countries start regulating fully autonomous cars.
- 2024: Cars start to be made which can only be operated autonomously. An autonomous-only car will be just about $4,000 more expensive than a manual-only car, and it will be much cheaper to insure.
- 2025: Carmakers expand their offering to autonomous trucks, vans, buses, etc. Each of those announcements is met with riots because it means that millions of people will be out of a job.
- 2026: Most of the developed countries have regulated fully-autonomous cars. Factoring in insurance, buying an autonomous car is cheaper than a manual car, meaning that manual cars will become what sport cars are today: a niche.
- 2027: A carmaker debuts a car with noise insulation, smart windows that can be made opaque, and a comfy bed.
- 2030: More than half of teenagers say that they aren’t going to learn how to drive.
- 2035: More than half of the new cars sold in Europe and North America have full autonomous capability. More than half of those cars don’t allow manual control at all.
I think trying to venture more into the future will just mean that I’m gonna say even more ridiculous stuff, so I’ll leave it there.
Reasoning for the timeline
About the Tesla “coast-to-coast autonomous drive”. I don’t doubt that they will make it. It’s a very bold marketing move, and it would be a significant blow for them if they don’t pull it off. But I think it will be in a very controlled environment. They have to go from Los Angeles to New York, so 99% of the trip will be on highways, a pretty safe environment for the current software. I’m sure that for the start and the end of the trip, they will carefully choose the route and the timing so there are no complex circumstances: no construction work, no pedestrian traffic, no policemen waving signs, no roundabouts, low traffic, etc. So the trip will succeed, but that won’t mean that the technology is there yet.
About the “Level 2” to “Level 3” transition, that’s a subtle line to cross. One could argue that some of the current systems are already being used as “Level 3” because people don’t heed the warnings and just watch a movie while their car is in “cruise control”. It’s not unreasonable to think that in 1 year those systems will allow it to do the same, but in a safer way.
For “Level 4” cars, I think 2019 is reasonable. Waymo has already done extensive tests, and Tesla is targeting 2018 for “Level 5” (which is insane, but I wouldn’t bet they are more than 2 years off a prediction so fundamental to their business model).
For the “autonomous option” price, the current Tesla Model S “fully autonomous package” option is $8,000, so my guess is that hardware prices are going to keep falling, and the rest of manufacturers will have to match those prices if they want to be competitive, even though they may barely cover costs (I’m talking, for example, about Waymo and their expensive LiDAR sensors). Eventually, when cars can be made with the “autonomous-only” option, prices will be reduced even further, because that means these elements will no longer be needed:
- Mirrors (which also offer considerable aerodynamic drag).
- Wheel & pedals.
- The whole cabin interior can be optimized for comfort or cost savings, since having seats facing front is no longer required.
It’s hard to predict what path insurance companies will take. Tesla has hinted that they will insure autonomous cars themselves, and I think that’s the approach that some carmakers will take: When you’re driving in fully autonomous mode, any accident that happens is either the fault of the other car, or a programming error of the autonomous software, so the automaker should cover it. I suspect insurance companies will offer “hybrid” plans, when they cover the driver fully in autonomous mode, but only partially (or not at all) in manual mode. Those plans will be much cheaper, since the accident rates of autonomous cars will be much lower.
Another hard to predict variable is regulations. Right now, even if your car was fully autonomous, you would still need to be behind the wheel, alert, sober, and without looking at your phone. Regulations will slowly allow multitasking when driving in autonomous mode. Texting while “driving”, or reading a book, would no longer be a crime. A few years after that, they will make the next jump and allow a fully driverless car. That would mean that a child could be “behind the wheel” (provided he can’t operate the manual controls), or a drunk person, or a person without a driver’s licence, or no one at all (imagine your car automatically picking your kids from soccer practice).
A hairy subject that will surely affect both regulations and insurance is the mortality rate of driverless cars passengers. Thousands of people die in the roads every month, and we’ve grown accustomed to this fact. Driverless cars will be safer by at least an order of magnitude, but they won’t be perfect. Maybe an unforeseen circumstance that could have been lethal at the hands of a human will be lethal for a driverless car too. Maybe a computer glitch will take a car through a cliff. Objectively, driverless cars will still be safer, but public opinion will be outraged by those few cases where they weren’t infallible. The big question is: When is it ethical to release a driverless car to market? Is it when it’s significantly safer than the average human driver? Than the average taxi driver? Than the best human drivers in the world? I can see why anyone would pick any of those options. A driverless car is essentially an automated taxi driver, so if you trust a random taxi driver to get you to your destination, why wouldn’t you trust a statistically better version of that? On the other hand, if you have a driverless system in your lab that’s significantly safer than the average human, and you don’t release it because you want it to be flawless, there will be a lot of people dying on the road every day that could be saved with your tech. On the other-other hand, if you release it and some small number of people die, the public will distrust autonomous cars and that will delay adoption on the long run (and even rollback regulations). People don’t understand statistics, and everyone thinks they’re better drivers than the average. This would be an interesting debate to have, and it will be interesting watching it unfold in the next few years, but I digress.
Consequences of the proliferation of driverless cars
Once the bill that allows driverless cars on the street passes, even if it’s in a single state or small country, there’s no turning back. Let me list the professions that will be affected:
Taxi drivers. Paying one human driver is more expensive that paying no human driver at all. I can guide you through the math of it if you want. You’ll be able to jump into a driverless taxi, select your destination from a touchscreen (or an app) and pay using your credit card, or PayPal, or whatever app the taxi company is using. Obviously, the minute it’s legal to do so, Uber and all the other ride-sharing companies will replace all their drivers with driverless cars. Some taxi drivers will remain, but their fares will be more expensive. They will become a progressively fading niche for people who don’t trust technology and prefer to pay more for “the human touch”. Eventually, the streets will become filled with driverless cars that will automatically “transform” into taxis when their owners don’t need them, making a few extra bucks for them.
Bus drivers. Even though tons of driverless taxis will be in the streets, I don’t believe that bus will become extinct. Bus drivers will, though, for the same reason as taxi drivers: economics.
Truck drivers. I think you’re starting to see a pattern. A driverless truck / van / car / bus can drive itself literally 24 hours a day, without getting distracted or demanding a pay. There’s no way to compete against that. That’s why I predict a lot of protests worldwide. The transportation industry is quite big, we’re talking millions of jobs in a single country, and they will eventually be lost (not without a nasty fight, if the current situation with taxi drivers against Uber is any indication).
Driving schools. If buying an autonomous car is cheaper (factoring in insurance), why would people learn to drive? In some countries, it’s a complicated and expensive process, and it will become a niche ability to have, with not much practical value. Sure, some people will learn to drive because they want to, they enjoy it or they can’t afford to replace their old car with one of those new robot cars, but an industry can’t be unaffected by a loss of say… 50% of their customer base. Let’s face it, most of us have a driver licence because we need it to go from one place to the other, not because we enjoy the thrill of gear-shifting when commuting.
Airlines. This may be a little far-fetched, but let’s make this back-of-the-envelope math: To catch a plane, you need to go to the airport (1 hour), go through security & boarding (1:30 hours), then when you arrive to the destination airport, you need to go to your destination (1 hour). So that means that for a short-distance flight of about 500 miles (1 hour of flight), the travel time is actually about 4:30 hours. You can do the same trip by car in 8 hours or less, with these added advantages:
- Much cheaper (compare the gas cost vs 2 taxis and 1 plane ticket).
- You have your car available at your destination.
- With an autonomous car, you can use a face mask and ear plugs, travel at night, and get your 8 hours of sleep in the car instead of going through a stressful airport experience.
So I think airlines are going to see reduced bookings on short-range flights of, say, 600 miles or less. Not the end of the world for them, that’s for sure, but they will surely feel it in their bottom line.
Enough with the apocalyptic prediction of millions of lost jobs. What will that mean for the average guy? The possibilities are endless. Once you accept that you can own a car that drives itself and you can control from an app, things start to look futuristic pretty fast. You will be able to:
- Step out of your car at your office’s door and let it find a parking spot for itself. Since it doesn’t have any rush, it can go to the suburbs or to special parking buildings for driverless cars, instead of paying an expensive urban parking space.
- The car knows your calendar (yay, apps!), so it knows where and when does it have to pick you when you finish work.
- Alternatively, you can configure the car to “become a taxi” the second you step out of it, so it will earn you a few bucks every day, instead of being in a parking space slacking off.
- Have the car go take your children to school or sports practice, and take them back. With a parental lock, of course, so they can’t alter the route.
- The camper’s dream: at night, get into your RV and configure where you want to be the next day. Go to sleep. In the morning’ you’re on your destination and you have the full day to enjoy it.
- Take a nap while you commute. Or read a book, have breakfast, check Twitter, read a book, etc. Make the windshield and windows opaque so nothing distracts you in your peripheral vision. Automakers will double down on comfort. Cup holders, a TV screen, outlets for small appliances like a toaster or a kettle, etc.
- Say goodbye to designated drivers! Everyone at the party can get wasted, and you won’t need to take expensive taxis home.
Just to reiterate, everything I’ve just listed it’s going to happen in about 5 years, 10 years tops. I’m not talking science-fiction here. But that’s just my vision of it. Let me know how I’m a lunatic in the comments!