It might still sound like science fiction, but driverless cars are here. Actually, they’ve been here for almost a decade. In 2004, the US Military put out a $1 million prize for anyone that could design a self-driving car around a course. Their goal was to encourage the development of new technology that could keep troops out of danger. While the initial prototypes were complete failures, the body of knowledge that grew out of those test projects has been extremely fruitful. Since 2009, Google has been testing self-driving cars on real roads. To date, these cars have logged over 3 million miles on real roads surrounded by human drivers. While driving those 3 million miles, only one collision was reported (when the car was rear-ended by a distracted human driver). For comparison, the accident rate on real roads by real people is about 6 accidents per 3 million miles. Other companies are making their own parallel efforts as well with promising success. In 2015, Nevada licensed autonomous semi-trucks. The state is also currently allowing the testing of self-driving “taxis.” We’ve already seen successful long-range test drives as well, driving 550 miles without a hand on the wheel.
This outsourcing of human tasks may sound intimidating. But remember that we have been using computer assisted transportation for decades. We’ve had automatic speed control in trains since 1788 and in cars since at least 1958. We’ve been allowing computers to help pilots maintain course since at least 1912. Computer assisted systems were integral to put men on the moon, ensuring the exact flight control patterns were followed.
The point is, as we continue to improve the technology that can assist us in getting where we need to go, it will improve our lives. Right now, we lose about 35,000 lives each year in the USA to vehicle accidents. Most of these are related to either speeding, impaired driving, or human error. This loss of life, personal injury, and property damage can be completely avoided in the future, with the help of computers.
Forecast
If you go shopping for a new car today, chances are that you will find even more computer assisted driving features than just cruise control and automatic gear shifting. You are likely to see options like “in-lane monitoring,” “park assist,” “automatic braking,” and “early collision detection” systems. All of these are improvements to prior versions of personal vehicles that improve safety. I expect that as we continue down this path, you will start seeing an “auto-pilot” option in personal use vehicle in the very near future. These cars will have advanced Lidar sensing and deep thinking computers. Eventually, this auto-pilot feature will become the default way we transport ourselves. And when that happens, it will disrupt the entire transportation system.
This forecast is consistent with demographic changes that we are seeing. The younger generation is not as interested in getting a driver’s license as their parents and grandparents were. In 1983, 92% of 20-24 year olds had a license. In 2014, that number fell to just 77%. And that trend appears to be continuing. What this means is that the future car buyers are likely to be drawn to the idea that they don’t have to drive. This is a stark departure from the previous three generations that enjoyed the freedom and control that driving afforded them. What this also means is that we are likely to see a long transition away from human drivers. As the previous generations cling to their desire to drive, the new autonomous vehicles will need to be able to share the road with them. This implies a 30 year period in which some demand for manual cars will remain, representing a smaller and smaller share of the cars on the road over that period.
The Shift
What’s even more interesting about this notion of driverless cars is the flow-through impacts they will have. In particular, think for a moment about the way you use your car. If you are like most people, you get up in the morning, drive yourself to work, park your car in a garage for 9 hours, drive home at night, and park your car in your garage for 14 hours. Sure, you probably make a run to the store, or movies, or whatever, once in while. And that freedom to know that you can go anywhere, at anytime, is valuable. You don’t have to wait on anyone else, you can just get in the car and go.
Let’s do some math. Let’s assume you have a 30 minute commute each work day and that you spend a total of 3 additional hours in your car each week running errands. That adds up to 8 hours per week in which your car is in use out of the 168 hours the week holds. That’s about a 5% utilization rate. That’s right, 95% of the time you hold ownership of your car is spent parked.
Let’s do some more math. My car payment is about $300 per month. I get about 30 miles per gallon and drive roughly 1,000 miles per month. That works out to about $100 per month is gas. My insurance is another $100 per month or so. I haven’t had a lot of maintenance issues yet, just a couple of oil changes per year and tire change-over in the spring and fall. That works out to roughly $400 per year, or $33 per month. Major maintenance costs tend to cost a lot, but only happen every few years (new tires, replacing the alternator, etc.). According to this blog, $400 per year on average is about right over the life of the car. I think that’s a little low, but let’s go with it. Finally, the cost to park that car adds up as well. I personally have a parking spot at work, but many people don’t. A casual look on the internet suggests that $5 per day is a good placeholder for parking. So, that’s another $125 per month if you only pay to park on workdays.
All told, that works out to around $668 per month, or $8,000 per year that we pay for transportation. Other sources calculate the average annual cost a little higher, around $8,700 per year. So, it appears we are in the ballpark with this rough estimate. Either way, if we assume the normal use of a vehicle is those 12,000 miles per year, the cost of a personally owned vehicle works out to somewhere around $0.65 -$0.75 per mile.
Because the average cost of paying someone else to drive you to work includes the cost of that driver’s time and a profit for the owner of the company, owning a car is usually cheaper than taking a taxi for most people. Of course taking a bus is cheaper still, but many people value the direct route to work, on their own schedule more than those savings. Most people prefer car ownership if they have the funds to afford one.
But imagine the situation where your car can drive itself. Perhaps then you would be willing to send your car out to give people rides while you were at work or asleep. This would result in two important facts. First, the utilization of your vehicle would increase, being parked less often. Second, you would collect fees for the use of your car, which would offset the cost of car ownership. This means that the cost of the car itself (the $300 per month I pay as a car payment and the $100 per month for insurance) gets spread across more miles, pushing down the average cost per mile driven. The variable costs (gas and maintenance) will increase as more miles are driven (my example works out to about $0.14 per mile), but you will recover greater than those costs (otherwise you shouldn’t do it). The parking cost may also reduce, depending on how much time your car is driving versus parking.
So, this means that the cost of transportation must decrease. And, because we no longer have to pay a driver for their time, the cost of hiring a ride is also pushed down (from somewhere around $2 per mile today to something closer to $0.20 per mile). When thousand of people are willing to put their car out there for other people to use (meaning your wait time would be short), and the cost of hiring these cars is being pushed down, the math related to car ownership starts to shift. So, the decline in driver’s licenses isn’t the only thing that will keep falling. Individual ownership of cars will also fade away.
Imagine
When we try to imagine the future, our imaginations anchor us to what we see today and extrapolate improvements from that starting place. So, when we think about self-driving cars, we are imagining the cars we drive today, but without the driver. In reality, we can’t know what changes will emerge. But, we can start to think about what a world without drivers looks like. When you need a gallon of milk from the store, you won’t have to send your husband, you can simply send your car. Need to pick up your kid from soccer practice? No need to leave work early, just send a car.
Think about what this will mean for parking lots, traffic police, insurance companies, and public transportation just to get started. Think about the design of a car that isn’t build for a driver. No need for front facing seats, mirrors, headlights, control panels, or even windows for that matter. And think about the flexibility that is provided by non-ownership. No need to drive a car designed for five people when you’re the only passenger. Just order that five person ride when you need it. Just need milk? Why drive a SUV when a small delivery vehicle can suffice? The point being, once we fully embrace computer assisted vehicles, the whole system will start to change. We will have fewer vehicle related injuries, more efficient transportation systems, reduced tasks to perform, and lower energy costs.
This will be a long, slow process, which is already a decade underway. I expect that within 5 years, the entire taxi industry in a few major cities will have converted to this automated ride sharing concept. Within 10 years, all new cars will have an “auto-pilot” option. Within 20 years, driving a car manually will be equivalent to using a dial-up modem today. This conversion will be slowest in the United States and fastest in emerging countries that won’t face the status quo resistance. And this technology will extend beyond cars. Airplanes, trains, barges, semi-trucks, tractors, tanks, helicopters, drones, and whatever new transportation system we build will all be autonomous. This is going to create great improvements in our quality of life, but will also upend the current economic systems built around transportation fuels. Stay tuned.
Note: I used my car payment as a convenient proxy for the cost of the vehicle itself. The technically correct way to determine this value is through a depreciation schedule, which is likely to generate a lower number than your car payment. In reality, my plug number is still low because I drive a lower cost car than the average person. I also made some other simplifying assumptions in order to be more readable. This is not a thorough economic analysis, but it gets to the same endpoint.