A future world full of driverless cars... seriously?!

A future world full of driverless cars... seriously?!
Will the reality match the hype that’s promised from a future with driverless cars? Credit: Shutterstock/Karsten Neglia

Driverless cars are an engineer's dream. At last, a technology that promises to remove the human factor from the traffic system.

It is humans, after all, whose contribute to 75% of road crashes, who introduce undesirable randomness into the mathematical simplicity of flows, and who have been characterised (somewhat tongue in cheek) as "monkey drivers" with slow reaction times and short attention spans.

If only we could eliminate the human factor, we would have cities teeming with safe, efficient cars whizzing us to our destinations. Right?

Wrong. For better or worse, as long as there are humans in the transport system we cannot ignore the human factor. To do so grossly overestimates the promised benefits of driverless cars and underestimates the negative impacts they will have on our traffic networks and society.

Think like a human

First, there are the immediate technological hurdles. At high speeds this is actually relatively straightforward as interactions on freeways are already effectively "vehicle-to-vehicle".

We are travelling too fast on a freeway to communicate at a human level, so we rely on infrastructure and to do much of the work for us, from using indicators to following dynamic signage. Removing human error is plausible and beneficial.

But all of that changes at low speeds, where drivers have to interact at a human level, such as when making eye contact with another driver, giving the nod to a pedestrian, or waving to a cyclist to let them go ahead.

How will an automated vehicle know if a pedestrian standing near the zebra crossing is waiting to cross or chatting on the phone? How will it process regional differences in body language, such as Google Car's confusion over a "track-standing" cyclist?

Google is already training its cars to recognise a cyclist's hand signals, but we still have a long way to go.

Similarly, without human gestures, how will the rest of us learn how to anticipate the actions of driverless cars? Recent research suggests that we don't yet know.

Making humans comply

One of the issues with the utopian vision promised by driverless cars – cities where parking is converted into parks, or intersections where traffic lights aren't even needed – is that it only works if 100% of the vehicle fleet is automated and individual ownership makes way for a fleet of shared pay-as-you-go taxis.

But how many people will actually opt in to this vision of the future? If you don't trust the technology, if you get motion sickness, if you enjoy driving classic cars (or motorbikes), or if you just don't like the idea of being driven by a car that always follows the speed limit and never jumps the queue, then a driverless car may not be for you.

It is no wonder that forecasts of the market penetration of driverless cars vary so enormously. For example, estimates from the Netherlands range from 7% to 61% of the vehicle fleet by 2050.

Even if we do reach 100% car automation, we still cannot ignore humans. Smart automated intersections promise to remove the need for traffic lights and allow twice as much traffic to use the roads.

But how will non-automated cyclists approach these intersections? How will pedestrians cross them?

We may reach a stage where the road safety benefits of driverless cars are so blatantly evident that non-automated cars are made illegal, and we wonder why humans were ever trusted to drive.

We are “monkey drivers.”

But until that day we will be living in a messy world of haves and have-nots with all the infrastructure required for both systems to run in parallel.

No more car ownership

Then there's the issue with sharing a driverless car fleet, with some claiming driverless cars will mean we move beyond individual ownership.

Car-sharing systems have existed for decades in the United States, yet fewer than 1% of Americans are members. Even optimistic estimates top out at 10% of the market.

Car-sharing has enormous potential in compact cities such as San Francisco or inner Sydney, where individual car ownership is expensive or impractical and many trips can be completed by public transit, cycling or walking.

But if you live in the suburbs or a rural area, if you have one or more child seats, if you store and carry goods in your car, if you want to have a say in the style of car you ride in, then it is unlikely that car-sharing will be economical or desirable for you.

Gaming the system

If driverless cars are instead owned by individuals, that opens the door to gaming the system in a way that is likely to erode the promised congestion-busting benefits.

Humans have an uncanny ability to make any system work for their individual benefit. When that happens, the congestion benefits promised by driverless cars are likely to be quickly undermined by human nature.

The small congestion benefits promised through freeway platooning and efficient intersections are likely to be quickly undermined by increased use of driverless cars.

It's also true that the more attractive you make travelling in driverless cars, the more people will do it. If you can catch up on emails during your hour-long drive, why bother to take the train? But some of the tactics that might remove the hassle from driverless travel could also worsen traffic.

Allowing driverless cars to run without passengers opens up an enormous potential for exploitation. Why pay for parking downtown when you can send your car back home to park (doubling the trips in peak hour in the process)?

Why bother to find a parking space at all if your car can circle the block by itself while you order a latte?

Changing society, one car at a time

The biggest changes to society expand far beyond individual drivers. The largest benefit, by far, is reducing the road toll, which costs Australian society A$27 billion per year. Thousands of deaths and serious injuries might be prevented through automation.

Yet this is not the only potential impact. Allowing the disabled, blind and unlicensed access to a driverless car will provide them with unprecedented freedom and mobility, but it will also increase cars on the road by 2-10%, once again eroding congestion benefits.

Driverless cars will also threaten the jobs of people who drive trucks, buses, taxis and Uber cars. In total, this is about 2.6% of the working population, according to the 2011 Australian Census.

Fewer crashes means fewer jobs in car repair and insurance, while compliant cars mean fewer parking tickets and speeding fines, reducing government revenue.

So despite all the hype, promise and predictions, no one really quite knows what the future of will look like. But as long as humans are leaving their homes, we cannot ignore the factor.


Explore further

How driverless cars and mathematics could spell the end of traffic jams

This article was originally published on The Conversation. Read the original article.
The Conversation

Citation: A future world full of driverless cars... seriously?! (2016, October 10) retrieved 20 August 2019 from https://phys.org/news/2016-10-future-world-full-driverless-cars.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.
14 shares

Feedback to editors

User comments

Oct 10, 2016
Human drivers are going to be a higher and higher percentage of all crashes as driverless cars take over the market. Insurance rates will skyrocket for anyone still trying to drive themselves until simple economics causes everyone but the ultra rich to adopt a driverless solution. I honestly have been looking forward to turning over the responsibility of driving to a machine and turn drive time, into gaming time. Something the size of a work van could contain a Holosuite with treadmill floor to let you get your morning run in to earn the Healthcare Dailies that keep your insurance cost as low as possible, while you are on the way into the office.

Oct 10, 2016
How will an automated vehicle know if a pedestrian standing near the zebra crossing is waiting to cross or chatting on the phone? How will it process regional differences in body language, such as Google Car's confusion over a "track-standing" cyclist?
Well how do we? Are you saying there is some sort of non-physical perception available to humans that machines can never have?

More importantly, how often does the average human driver get it wrong?

I'm sorry I rarely declare an entire article rubbish but today's a new day.

One thing to consider is that humans will be adapting to these new technologies as well. A pedestrian knows better than to cross in front of a train. Most do at any rate.

Oct 10, 2016
I don't see any situation that a driver AI can't learn eventually and handle it as well or better than a human driver. Yes, it will take a long time to cover all situations, but as long as the humans interacting with those systems don't start doing stupid things in order to confound them (such as the 'track-standing' cyclist mentioned above), they'll work just fine.

Personally I can do very well without other people hand-signing me behind colored windows that I can't see through or that reflect the sun in such a way that it's impossible to see inside. Likewise, I can do without pedestrians standing next to a zebra crossing waving me through. Either they want to cross or not; if the former, they should take the hint when I'm slowing down. If the latter, they shouldn't stand right at that crossing.

Oct 10, 2016
Or we could just sidestep the whole problem and finally get some friggin' flying cars (hexa- or octocopters preferrably). Then it's just an issue of pathing which computers can solve easily without any human input.

I honestly have been looking forward to turning over the responsibility of driving (to a machine) and turn drive time, into gaming time.
(brackets mine)
Many people already have. It's called: taking the bus and playing on your smartphone ;-)


Oct 17, 2016
Human drivers are going to be a higher and higher percentage of all crashes as driverless cars take over the market.


That's assuming the driverless cars never get it wrong. Human drivers are already blamed for nearly every crash, so it's rather the opposite that will happen - the more humans are not driving, the higher the percentage of crashes become due to hardware faults and computer errors.

However, I'm sure they will find a way to pin the fault on people somehow, because if they don't then the company that made the machine becomes liable. Manufacturers can never admit that their products or procedures are unsafe - that would affect sales.

That's already the case with airplane accidents, where the ultimate fault upon examination is always "pilot error" even when the computers made the error and the pilot had no real means to rescue the situation without supernatural skills and awareness. Last one to touch the stick gets the blame.

Oct 17, 2016
. If you can catch up on emails during your hour-long drive, why bother to take the train?


Why indeed, considering the train will often be less efficient in getting you from A to B anyhow, taking you considerably longer in both distance and time and usually requiring you to take two different cars as well as the train to get where you're going.

Tracks don't go from door to door.

Or we could just sidestep the whole problem and finally get some friggin' flying cars (hexa- or octocopters preferrably). Then it's just an issue of pathing which computers can solve easily without any human input.


And then a magnetic storm downs the entire fleet, hopefully not while they're flying.

That's also a problem for self-driving cars that depend on GPS.

Oct 20, 2016
"So despite all the hype, promise and predictions, no one really quite knows what the future of driverless cars will look like. But as long as humans are leaving their homes, we cannot ignore the human factor."

Funny as this articles last paragraph essentially nullifies the requirement of the entirety of the article before it.

Also put this one under opinion piece.

It's a con that there will be less auto repairs or injuries to support that financial infrastructure?

So you place money over life essentially. Got it.

How many driver-less cars would drive under the influence in a year?

Choose to stunt?

Disregard traffic signals?

But I guess the "Human aspect" explains a lot about the condition we find humanity in today.

Oct 20, 2016
That's already the case with airplane accidents, where the ultimate fault upon examination is always "pilot error" even when the computers made the error and the pilot had no real means to rescue the situation without supernatural skills and awareness. Last one to touch the stick gets the blame.


Ya do you know about an aircraft manufacturer called McDonnell Douglas at all? Look up Case No. CV-74-2007-PH.

It's history makes this part of your comment "QUANTIFIABLY WRONG"!

Also how many times will self driving vehicles speed?

I mean it's only the police and highway transportation administrations around the globe that all agree speed is the biggest killer.

Got an answer for that?

Oct 21, 2016
It's history makes this part of your comment "QUANTIFIABLY WRONG"!


The case you present, Saxton vs. McDonnell or "Civil Liability for Causing or Failing to Prevent Suicide" seems to be entirely irrelevant to the question.

Also how many times will self driving vehicles speed?


They do. They're programmed to drive with the flow, because sticking to the exact speed limit causes unwanted oscillations in the traffic flow.

I mean it's only the police and highway transportation administrations around the globe that all agree speed is the biggest killer.

Got an answer for that?


That's like saying "guns kill" when it's the person pulling the trigger who kills. Speed itself doesn't kill - speed in the wrong place kills, and automated vehicles are perfectly capable, or even prone to making those sort of misjudgements.

Again, the computer is not smart. It's currently very dumb, deaf and almost blind.

Oct 21, 2016
The city's revenue departments are not going to like completely driverless traffic.

Oct 22, 2016
City people pipe dreams.

Oct 22, 2016
Perhaps now would be a good time to start thinking about how to safeguard them from hack attacks.

Oct 22, 2016
I would like to see the proponents of self-driving vehicles demonstrate that these vehicles can detect and respond appropriately to motorcycles in a variety of complex real-life situations. If they can do this then I will be their biggest fan. Actual human drivers are generally terrible at dealing with motorcycles. If they can not demonstrate the above, I cannot support driverless technology.

If they can demonstrate real competence with motorcycles, it would save many lives and countless needless injuries. But curiously, I've seen nothing on the matter. How come?

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more