There were 5.6M traffic accidents in the USA in 2013. Self-driving cars from Google and trucks from Freightliner (owned by Mercedes-Benz) have been in the news recently. Self-Driving Vehicles (SDVs) will reduce traffic accidents, transportation costs, transit times, and road congestion. (See Google Self-Driving Car Project.)
Uber is valued at $41B today (and rumors of a $50B valuation are circulating) because investors in this still privately-held company are betting on Uber being a leader in SDVs.
A recent report estimates there will be 10 million SDVs by 2020.
(Note: While this report estimates various flavors of self-driving, in the rest of this post I am focused on 100% autonomous vehicles: no human driver.)
The change-over to self-driving cars, buses, and trucks will cause dramatic changes in many different industries. Government and school bus fleets will be history, the arguments for rail lines (light or otherwise) will crumble, and human drivers of all of these vehicles will be out of a job.
Why do I own a car?
I own a car because it gives me the freedom and flexibility to go where I want when I want. But that freedom has a big price.
My late-model Mercedes-Benz convertible is comfortable, stylish, and gets to 60 mph very rapidly (and its twin-turbo V8 gasoline engine sucks fuel like a 1970s automobile from Detroit). But it is very expensive to own when you consider all of the costs and divide those by miles driven.
Not only did I have to purchase my car, but I also have to pay for automobile insurance, any damages or maintenance, new tires and brake pads occasionally, and when I eventually sell it (to get another car), its value will have depreciated by 50-70%.
I have to put up with traffic, I have to spend time driving (225 hours over the last 5,130 miles traveled), and I have to pay for parking most of the time when I am away from home. [Of course I also pay for parking at home by virtue of having built a garage and paying property taxes, and if I lived in a condominium in big city like NYC or Chicago, I would likely pay even more for a parking spot at home.]
I think it is very likely that SDVs will make enough progress in the next 5-10 years that this may be this last car I own.
An SDV will be better than my convertible
I will save a lot of money (as detailed above) if I don’t own a car. But of course I will have to pay Uber, Lyft, or other companies for every trip I take.
Instead of boring you with a detailed cost analysis and comparison (I am confident Uber management and investors have already done this math), let me suggest all the reasons why an SDV will be both less expensive and more convenient for me than owning a car.
Less expensive:
- An SDV needs to be safe, reliable, and clean.
- But since I do not own it, an SDV does not need leather seats, 19″ alloy wheels, low-profile sport tires, carbon fiber trim, a sexy body shape, Obsidian Black Metallic paint, or any of the myriad other cosmetic features that automobile manufactures provide to appeal to individual consumers.
- An SDV does not need any components required by human drivers: mirrors, instrument panel, dashboard, steering wheel, accelerator and brake pedals, gear selector, parking brake, spare tire, car jack, audio controls, interior climate controls, owner’s manual, …
- An SDV might have audio speakers and an LCD display, but more likely would rely upon the passenger’s smart phone to control the interior climate.
- An SDV will never attempt to surpass the laws of physics, so it will have a modest power plant (gasoline or electric), smaller brakes (computers have much faster reaction times, never get distracted), and much less expensive wheels and tires (only ride comfort and traction will matter).
- And SDV will not have a “lead foot”, so tire wear will be reduced and fuel efficiency will be improved.
- Without the need for a human driver to see out, the SDV will have a much more aerodynamic shape, which will be much less expensive to build and will increase fuel efficiency.
- Most personal cars spend 99% of their time parked, losing value. But an SDV can be in motion 24 hours a day, 365 days a year (except for refueling, cleaning, maintenance, and repairs).
More convenient:
- I never have to worry about refueling/recharging, washing the exterior or cleaning the interior, windshield wiper fluid, changing the oil, changing the radiator fluid, summer vs. winter tires, buying new tires, …
- I never have to spend time and money on parking — neither at home nor away from home.
- I suddenly have more free time for email, phone calls, work, or to just relax. And texting while being a passenger is safe. 😉
- I don’t have to worry about navigating between destinations, estimating travel time, taking a wrong turn, running into unexpected traffic or construction.
- If I want to haul a lot of people or a lot of stuff or head up into the mountains to ski or snowshoe, I do not need to own multiple vehicles.
- If I still had children in K-12 schools, I would not have to drive them around: to school, to soccer games, to after-school activities (but of course I would not be able to listen in on their conversations with their friends).
Uber in 2020 vs. Taxis and Buses
Uber had 162,037 active drivers in the USA in December, 2014, so it likely has north of 250,000 drivers today around the world. Uber is not only offering a valuable, convenient service today, but it is also collecting a lot of data from all of those trips that will allow Uber to optimize its fleet of SDVs.
Declining Price of SDVs
Self-driving technology adds $75,000 to the cost of a car today, but one recent study predicts that self-driving technology will fall to $7,000-$10,000 by 2025.
A basic Honda Fit — an award-winning small four person sedan — lists for $16,000 today. If you consider all the ways I listed above to reduce the cost of an SDV, you could probably halve the cost. The self-driving technology prices are falling rapidly, so you can estimate that an SDV Honda “Mini-Fit” might cost say $30,000 in a few years.
Uber will crush Taxis
If you have any experience with the yellow cab taxis in New York City (I have been to NYC 10+ times in the past 3 years), you know the cars are often smelly, unclean, have bad shocks, and the drivers subject your body to all sorts of roller coaster G-forces and visual upsets.
An Uber SDV will be cleaner, safer, more reliable, and less expensive:
- Taxi drivers are only allowed to drive at most a 12 hour shift.
- So taxi drivers naturally pick the 12 hours span that is most profitable for them, roughly 4:00 AM to 4:00 PM.
- Which means that between 3:30 PM and 5:30 PM, it is very difficult to find a taxi in NYC!
- But the Uber SDV will be constantly on call, and with the vast trip history Uber has accumulated, it can predict where and when future trips will be and “stage” SDVs nearby to respond to demand.
- With no human driver, there are no food breaks, coffee breaks, or bathroom breaks, and of course no human driver to pay.
- Uber will self-insure, as accident rates will plummet with no fallible human driver: (In six years and 1.7 million driven miles, Google SDVs only experienced 11 accidents.)
- As above, SDVs will have lower operating and maintenance costs because the software will drive more gently.
- SDCs trips will also be faster and more efficient, as the software coordination via the cloud will ensure optimal routing and avoidance of traffic jams.
- (Note: All of these apply equally to limousine car services.)
Uber will crush Buses
The vat majority of bus systems in the USA are government-run monopolies. With all of the features and inefficiencies you would expect:
- Fixed routes and fixed schedules.
- Inconsistent adherence to schedules — sometimes due to traffic, sometimes due to a lack of customer focus.
- High capital equipment costs due to government contracts.
- High labor costs: the average bus driver makes $63,040 in San Jose, CA (2013), and in some cities transit workers can earn more than $100,000 (2011).
Uber has been testing UberPool since at least August, 2014. As the name implies, this is a multi-rider “carpool” service. You share your Uber with one or more other riders, and it may not take the most direct route to your destination, but you pay less than you would for a solo trip.
Naturally, time, money, and comfort trade off in transportation. In the Uber smart phone app of 2020, you will select how many people in your party, how much luggage/cargo, and your starting and ending destinations, Uber will display a real-time list of price and travel time choices. You’ll pay the most for a large vehicle that travels directly to your destination. And you’ll pay the least for a multi-person vehicle that is projected to make (say) 3 stops on the way to your destination and take (say) twice as long.
Since Uber will dynamically optimize routing and pricing for UberPool, it will have both a convenience and price advantage over existing government bus systems. Similarly, public and private school buses will not be competitive, nor will charter buses. Uber will own and manage a fleet of Self-Driving Buses that can be deployed in specific situations (getting fans to/from a football game, a marathon start line, etc.).
Vast Positive Benefits of Self-Driving Technology
As more self-driving vehicles appear on our roads, many additional benefits will arise:
- There were 5,687,000 police-reported motor vehicle accidents in 2013: 30,057 fatalities, 1,591,000 injuries, 4,066,000 property damage only. As self-driving vehicles take a larger share of trips, these numbers will begin to fall dramatically.
- More road capacity for free in crowded cities, as the reduced need for parking will lead to the elimination of on-street, curbside parking.
- With more road capacity and the cloud-based routing of self-driving vehicles, traffic jams will become a distant memory, reducing transit time and fuel costs per mile driven.
- The Cloud Vehicle Coordination System (CVCS) will be able to track pedestrians (via their smart phones, of course), preventing vehicles from striking pedestrians.
- Speed limits will be raised on highways and many multi-lane streets, as the SDVs and the CVCS will be able to operate safely at higher speeds.
- Dynamic one way roads will be possible, able to respond flexibly to travel demands and increasing the available throughput on existing roads. (Today, there are “reversible” lanes in some cities, but they can take hours to switch directions, and follow a mostly static schedule).
- No more stop signs and red lights: with inter-vehicle coordination, SDVs will automatically interleave at intersections.
- Real estate devoted to below ground, surface, and above ground parking can be used for other purposes.
Disruptions caused by Self-Driving Technology
Naturally all this technology change will disrupt many existing industries and jobs:
- There will be a declining demand for personal vehicles, and the average selling price of self-driving cars will be dramatically lower than current average vehicle prices.
- There will be a declining demand for taxi drivers, chauffeurs, bus drivers, and truck drivers.
- There will be a declining demand for city, county, and state policemen to monitor traffic and respond to traffic accidents.
- There will be a declining demand for automobile mechanics.
- There will be declining demand for automobile insurance.
- There will be declining demand for automobile loans.
- Ignoring the effects of all-electric vehicles, there will be a declining demand for automobile fueling/service stations and automobile car washing stations.
- There will be a declining demand for traffic signs and signals.
A safer, faster, more comfortable, and less expensive future is coming our way.
Ben Slivka 
I agree with all of this 100%.
But as an interesting counterpoint, I recently had dinner with a friend who was many years at Car And Driver in an editorial and executive capacity. He was actually very bearish on SDCs for reasons that had me chagrined.
1) He was concerned that the regulatory/legislative hurdles would take decades to overcome. I made (what I thought were) compelling arguments otherwise. But still, if he’s more right than I am, that is a concern.
2) WAY more interesting, he suggested that liability issues were such a hurdle as to seem to him to make it almost a complete non-starter. He pointed out that, currently, because an individual controls the car, the liability is on the individual for whatever goes wrong. But in a SDC universe, any mishap, or accident, any anything, would be on the COMPANY and/or whoever is responsible for the software. Currently, Ford is only sued when there is something negligent in their design/manufacturing. But that would not be so in an SDC world. Any malfunction out in the real world would now open up an SDC manufacturer/operator to liability. And it’s not just if the software went wonky. Yes, we saw those recent stats about with the Google cars the accidents were few and far between and none of them were the cars’ fault. But he said, with a real would incident, EVERY ONE has the potential to be adjudicated. So literally, every real would incident has the potential of a lawsuit. Even if you won ever case, think of the cost! He wondered what company would be willing to stick their neck out for that.
3) He claims (again, I’m trusting his industry knowledge) that ANY adverse weather makes fully autonomous driving impossible at this point. Snow. Rain. Etc. So, yes, you can do your email while you get somewhere, but you still need to be available to take over if things go south, at least with current tech. So, the fully robot cars that can eliminate the need for busses? According to him, we’re not there for a while.
Again, I’m with you in pining away for a SDC future, so this is very much devil’s advocate sort of thing. And he’s a auto industry guy, so I don’t know if that colors your opinion of what he said or not. But the questions he’s raised… I don’t hear them in a lot of other places. Made me think.
1) Regulatory: I agree, the entrenched interests can lobby hard to get Government to protect them from Self-Driving Cars. But I think the economic and safety advantages will overwhelm the protectionists.
2) Liability: This is easy. SDCs are going to be way safer, dramatically reducing the 5.6M annual auto accidents in the USA (30K deaths, 1.1M injuries, rest property damage). I’ll write a separate post on how video recording with chain-of-custody will dramatically reduce disputes.
3) Human beings manage to drive (often poorly) in bad weather. New and improved sensors will meet or beat human visual acuity.
4) People in the USA still own horses, but rarely use them for transportation; human-operated cars will suffer this same fate.
Thank you!
–Ben
My concerns:
1. Computer glitches.
2. Malicious hackers.
3. Do-it-yourself hackers who will alter the vehicle’s software. i.e. to get to their destination more quickly, snag a better parking space, etc.
John
1) Computer glitches will be repaired. Something that can be done reliably and systematically. Unlike human drivers! Coincidentally, I was side-swiped in Evanston on Saturday morning. The young woman driving the other car was speeding (according to two separate witnesses) and she tried to pass me on my right as I was turning right into an angled parking spot. There were parked cars in front of her, so there was no where for her to go. She likely caused $3-4K of damage to my rental car and roughly the same amount of damage to her car. We exchanged information and I took a video of the accident scene and some still shots. But we did not call the police as I had a graduation ceremony to attend. So our ~$6-8K property damage accident will not be included in the 2015 police automobile accident statistics.
2) Malicious Hackers are not going away. So these computer systems will become hardened to prevent all but the most determined (likely state-sponsored) hackers. At some point, it would be cheaper to deploy human beings to cause mischief.
3) DIY Hacking will be pretty difficult (as above) and offer few advantages. The vast majority of SDCs will be owned by fleet operators like Uber, Lyft, etc. So passengers won’t worry about parking — the SDCs are just circulating, and on-street parking and parking lots will disappear. The cloud routing systems and individual SDC control systems will be optimized to minimize transit time — hacking wouldn’t be able to improve upon that.
I just hope that the software in SDCs works better than that in my iPad and iPhone, which are crashing at least every other day. Admittedly, the computer in my KIA has been working flawlessly for4 years, but it doesn’t rely on much external input. That is, external to the vehicle.
Your iPad and iPhone permit third-party applications to be installed and run in the background. Which opens you up to unexpected behavior. The SDCs will not allow random third party software installations, so they’ll be much more reliable. Again, the quality bar for SDCs is human drivers — either driving themselves, or paid to drive others. How many times have you been in a taxi when the check engine warning light and low brake pad warning light are *not* illuminated? 😉
This is a thorough analysis and I agree mostly. I am maybe less impressed than you by uber’s competitive advantage in this market, but they certainly have some.
My biggest concern is with the consequences of making travel by car cheaper, safer, faster, more pleasant, and more convenient. All these changes will make people willing to travel by car much more. That means people will tend to live further from their workplaces, and go more places (why not travel across town in rush hour just to see a friend, if I can read or nap while I’m doing that?), so the total amount of travel is likely to increase greatly, using more energy, causing more environmental damage, and requiring more roads. Sure, more efficient and electric cars will somewhat counter these effects, and SDVs will take up less space on the road, and there are benefits to people living where they want and traveling more, but the net effects could be bad.
Hi Bruce!
Thank you for your kind remarks and thoughtful comments!
Please consider:
a) “Rush-hour” will be a thing of the past as self-driving fleets come online. For all the reasons described, traffic will flow smoothly and quickly — even more quickly when we remove speed limits and red lights, since coordinated self-driving cars won’t need those.
b) I tried to explain while we’ll get way more capacity out of existing roads — the combination of no on-street parking, optimized ride-sharing, inter-car coordination, no traffic jams, etc., will all result in many more people-per-hour moved on current roads.
c) Even if you assume minimal change in how vehicles are powered, self-driving cars will transport many more people per gallon per mile than existing cars because: 1) they will have smaller engines driven in a more fuel-efficient fashion (no hard acceleration, no stop-and-go traffic); 2) vehicles will be “right-sized” for traffic — 1, 2, 4, 8, and 16 person vehicles, and no large diesel public buses riding scheduled routes with only a handful of passengers at off-hours.
d) Travel is a good thing. I was in rural Burma last January, where it takes a woman six hours to haul 50 pounds of produce on her back down the mountain to a market, sell the produce, buy 50 pounds of other stuff, and haul that back up the mountain on her back. Fast travel makes it easier for a kid to attend the best school for her, or a consultant to visit more clients, or a restaurant to attract more customers.
When I add all these positive effects up, I predict that per capita energy consumption will decrease while productivity and per capita GDP will increase.
2017-12-06 funny on self-driving care milestones: https://xkcd.com/1925/