Study: stopping at red lights exposes drivers to high levels of particulates
SLAC X-ray laser provides first glimpse of a chemical bond being born; implications for more efficient chemistry

Study: fully self-driving cars could result in fewer cars, but more miles driven per car

Autonomous vehicles (completely self-driving, level 4) may reduce the number of vehicles a family needs, but may lead to an increase in total miles driven per vehicle, according to a new analysis by researchers at the University of Michigan Transportation Research Institute.

UMTRI researchers Brandon Schoettle and Michael Sivak examined the 2009 National Household Travel Survey (NHTS) data set, which contains detailed information about each trip made by a person within a selected household, including the exact start and stop times of each trip. They found a general lack of “trip overlap” between drivers within a majority of households based on vehicle sharing. In other words, families rarely use more than one vehicle at a time.

The study is based on sharing of completely self-driving vehicles that employ a “return-to-home” mode, acting as a form of shared family or household vehicle.

To illustrate the self-driving vehicle sharing concept, consider the following example. A household has two drivers, A and B. Driver A requires a vehicle for commuting to work by 8:00 a.m. and home again at 5:00 p.m. Driver B normally runs errands during the day while driver A is at work, returning home after each errand. The level 4 self-driving vehicle “return-to-home” mode would involve the vehicle performing the following basic actions:

  1. Drop driver A off at work by 8:00 a.m.
  2. Return home to driver B for an errand from 10:00 a.m. to 11:00 a.m.
  3. Take driver B on a second errand from 2:00 p.m. to 3:00 p.m.
  4. Return to work location of driver A by 5:00 p.m. for commute home.
—Schoettle and Sivak

In the most extreme scenario, self-driving vehicles could cut average ownership rates of vehicles by 43%—from an average of 2.1 vehicles to 1.2 vehicles per household, the researchers say.

However, the shift could result in a 75% increase in individual vehicle usage—from 11,661 to 20,406 annual miles per vehicle (this increase in mileage does not include additional miles that would be generated during each return-to-home trip).

Schoettle and Sivak found that, on an average day, nearly 84% of households had no trips that overlapped or conflicted. Just under 15% of households had two drivers and less than 2% had three drivers with overlapping trips that created a conflict.

The researchers say their results represent strictly an upper-bound approximation of the maximum possible effects of self-driving vehicles on reductions in household vehicle ownership, given several unknowns: sufficient gaps between trips, acceptance and adoption of autonomous vehicles and possible vehicle-sharing strategies within households.

In the hypothetical scenario outlined in this analysis, the potential reduction in vehicle ownership per household would include a correspondingly large increase in vehicle usage, consequently increasing wear-and-tear and required maintenance frequency, while reducing the average vehicle life span (in total years on the road). Given the current average on-road vehicle age of 11.4 years and the assumption that the underlying scrappage rate are largely functions of overall wear-and- tear and total mileage, then a 75% increase in annual mileage per vehicle could reduce this average on-road age to approximately 6.5 years.

… One possible silver lining to decreased vehicle life span (in years) involves the more rapid introduction of new technology into the on-road fleet, and the consequent benefits for road safety. In the hypothetical scenario we describe in this analysis, with an average turnover rate that is 75% faster for these vehicles, the rate at which new self- driving technology would be replaced or updated in the on-road fleet would nearly double (versus current average vehicle age and annual usage rates).Schoettle and Sivak


  • Brandon Schoettle and Michael Sivak (2015) “Potential Impact of Self-Driving Vehicles on Household Vehicle Demand and Usage” UMTRI-2015-3



I honestly don't think there would be as much of a concern for muggings or theft in general with the autonomous taxi service...

1 the autonomous taxi is not going to stop for strangers unless its hailed and paid for, (not many will commit those types of crimes if almost assuredly they will get caught) (The cab could be made to lock and drive the perpetrator to the police station if necessary) Again, crime is the anomaly not the norm.

2 cameras will easily deter most assailants.

3 you could have help/distress buttons or a person sitting at a bank of computer monitors over seeing 10 cabs at once...

4 have the option for ride sharing or not (again, I wanted to add incentive, take the total fare and knock 10% off of the total price for every rider, then divide that amongst the passengers)

I think if there is incentives, and that if it is cheaper to use a taxi then own a vehicle, there will be significantly less cars on the road.

Also, you have to figure that there are waves of traffic, rush-hours are a huge influx of cars, but most other times traffic is much less.

Imagine if you are at work, and you and 3 coworkers (you've never met) hop into a cab and it takes you all home for less than driving your own vehicle. You'd never put this carpool together on your own, nor would you know where your co-workers lived to associate the distance/time/cost benefit. I honestly think that this is an incredible untapped opportunity.

You could also make it mandatory for a background check to ride in a taxi with others, felons would have to ride alone.

I don't think smaller cars are the answer, if everyone has a personal sized vehicles there would be longer streams of traffic other than on those special laned roads, imagine exit lanes or any stop light traffic...
Also the narrow wheel base could pose issues with stability and traversing different unexpected obstacles


As was said before there does not have to be any stability issues with a narrow car. You could, as Henrik & Alex C, suggested, put the batteries under the floor: They did this in the Tango and it has the same rollover threshold, 56°, as a 911 Porsche.

Another approach is to design the car to tilt into corners.


BTW, my fav tilting car was GM's "lean Machine."


I meant Renault Twizy EV when I wrote Renault Twingo.


If you can design a narrow, tilting car that passes federal safety standards and actually scoring decently in a insurance crash test rating more power to you.

Almost all of the examples are trikes... they follow a different set of rules much like motorcycles. Not a bad thing just apples to oranges.

If you can fit them side-by side in existing lanes fine, but I wouldn't imagine infrastructure change to accommodate more personal vehicles, when most problems can be alleviated by less cars and more people per vehicle.

Roger Pham

Good point, ai vin, that platooning can increase road capacity.
However, these must be done by vehicles equipped for the task, with the right electronics and communication equipments. Perhaps HOV lane can and should require future vehicles equipped for platooning. The computerized jitney vans would be ideally suited for the task, due to the presence of telecommunication equipments on board, already necessary for computerized jitney service for routing, pickup and delivery. Imagine a 10-12-passengers jitney van capable of platooning in the freeway HOV lanes, imagine how much more roadway can be saved, perhaps by 50-100 folds.

On the other hand, asking all cars on the road to be equipped with platooning equipments would be asking for too much. These equipments and platooning cars, if malfunction, can cause much more widespread damages like large chain collision, than 2-3-vehicles accidents typical of now. Thus, commercial vehicles under strict service and maintenance schedule of a large corporation can have much greater reliability than an average private vehicle on the road today!

Roger Pham

People prefers big cars and SUV's for safety over regular-size vehicles. For this reason, they will more likely choose large 10-12-seat jitney van with 4 rows of seat rather a regular car. People who prefer to sit alone can pay double or triple fares and occupy a whole row of seat.

The addition of a human attendant paid at $10 per hour would only add $0.03 additional cost per passenger mile, a negligible expense, consider the fact that personal car owners paid on average 69 cents per mile in 2012.
Until Walmart will start replacing greeters and checkers with robots, then employing a human attendant/driver instead of robots, robocops and robochauffeurs etc would be a more economical thing to do. Who would know more than Walmart than the economics of these issues?


How the authors can come up with a study that seriously contemplates having empty vehicles making a full additional round-trip between home and work each day is beyond comprehension.

That will nearly double the total distance driven, the energy use, and the amount of traffic, for a benefit that can easily be had by collective transportation or sharing vehicles, rather than having privately assigned vehicles.

Conclusion: Idiocy is alive and well in American car culture, not withstanding recent trends of driving less.

Now, before anyone starts arguing that nearly doubling distance driven, energy used, and traffic congestion can be alleviated by technology, I will say one thing: Such technology can and should instead be used to keep constant, or reduce, the distance driven, reduce energy consumption, and reduce congestion.

It ought to be pretty clear what the correct solution is. And it does not involve vehicles returning home after the work commute.

Anything that a car returning home can do, a car that is not constrained to return home, can do better.


I think you're missing the point Jus7tme. The self driving car would only have to return home >>>IF<<< another driver in the household needed it.


ai vin, you cannot be serious. Of course I know that. The article is about the supposed benefits of having cars returning home after the commute FOR USE BY A 2ND USER, and then return to work for a the evening commute.

Are you just being a troll or what? How about instead you respond to what I actually wrote, instead of trying to mis-characterize it.


Well, seeing as how you asked so nicely... ::rolling eyes::


Uh, hello? You talk down to me as if I am a misguided child, and mischaracterize what I said. And then **I** am supposed to ask nicely that you reconsider what you wrote? I think rather that you should apologize.


Just get over yourself and move on, I have.

The comments to this entry are closed.