Fully self-driving cars expected by 2030, says forecast — UPDATE
January 3, 2014
(Credit: Google)
Self-driving cars (SDC) that include driver control are expected to hit highways around the globe before 2025 and self-driving “only” cars (only the car drives) are anticipated around 2030, according to an emerging technologies study on Autonomous Cars from IHS Automotive.
In the study, “Emerging Technologies: Autonomous Cars — Not If, But When,” IHS Automotive forecasts total worldwide sales of self-driving cars will grow from nearly 230 thousand in 2025 to 11.8 million in 2035 — 7 million SDCs with both driver control and autonomous control and 4.8 million that have only autonomous control.
In all, there should be nearly 54 million self-driving cars in use globally by 2035.
The study anticipates that nearly all of the vehicles in use are likely to be self-driving cars or self-driving commercial vehicles sometime after 2050.
The price premium for the SDC electronics technology will add between $7,000 and $10,000 to a car’s sticker price in 2025, a figure that will drop to around $5,000 in 2030 and about $3,000 in 2035 when no driver controls are available.
Benefits of self-driving cars
“There are several benefits from self-driving cars to society, drivers and pedestrians,” says Egil Juliussen, principal analyst for infotainment and autonomous driver assisted systems at IHS Automotive. Juliussen co-authored the study with IHS Automotive senior ADAS analyst Jeremy Carlson.
“Accident rates will plunge to near zero for SDCs, although other cars will crash into SDCs; but as the market share of SDCs on the highway grows, overall accident rates will decline steadily,” Juliussen says. “Traffic congestion and air pollution per car should also decline because SDCs can be programmed to be more efficient in their driving patterns.”
The study also notes some potential barriers to SDC deployment and two major technology risks: software reliability and cyber security. The barriers include implementation of a legal framework for self-driving cars and establishment of government rules and regulations.
Autonomous cars by 2020
Several automakers have said publicly they will have autonomous cars by 2020, or earlier. Autonomous car technology is already affecting driver assist systems such as adaptive cruise control, lane keep assist, and collision mitigating brake systems.
Additionally, the IHS study says the first group of autonomous cars will have so-called Level 3 capability — limited self-driving that enables the driver to cede full control of all safety-critical functions under certain traffic and environmental conditions and includes auto pilot for highway travel and parking.
Coming later in the decade will be SDCs with Level 4 capability — self-driving but with human controls.
North America is forecasted to account for 29 percent of worldwide sales of self-driving cars with human controls (level 4) and self-driving only cars (level 5) in 2035, or nearly 3.5 million vehicles. China will capture the second largest share at 24 percent, or more than 2.8 million units, while Western Europe will account for 20 percent of the total, 2.4 million vehicles.
UPDATE Jan 3, 2014
KurzweilAI asked Dr. Egil Juliussen, principal analyst for infotainment and autonomous driver assisted systems at IHS Automotive, to comment on the following questions.
How do these forecasts on on self-driving cars compare to other ones?
EG: I am not aware of any other forecasts on self-driving cars. It is a little early to make such forecasts. This was done as a part of a large multi-client study called “New Urban Mobility” that looked at the major aspects of the auto industry until 2035. It became clear that self-driving cars would have a growing impact on the auto industry and I spent several months researching the technology and other elements of self-driving cars. The report that was released was extracted from this project.
Is Google in agreement with the dates? I have the impression that they are anticipating earlier dates for level 5 [fully self-driving].
EG: I don’t know if Google agrees. Hopefully they will buy the report and give us some feedback. Google is ahead of the auto OEMs today, and is likely to be a technology supplier at some point. The Level 5 or self-driving-only car (without driver control) has not been discussed much and we believe it will come 5 years later than the SDC [self-driving car] that can also be controlled by a driver. I would think that Google will be thinking about this too, but they have not said anything about this topic.
What about highway and road redesign to facilitate level 4 and 5? Is it necessary and if so, for what types of roads?
We did not address this topic, but it is an interesting question. The SDC has to be designed to drive on existing road infrastructure. Over time it is possible that the road infrastructure may be modified, but that is a long time off.
References:
- IHS Automotive, Emerging Technologies: Autonomous Cars - Not If, But When, 2014
Comments (50)
by cameronarndt
An important aspect that I think should be added to the conversation is the possibility for retrofitting existing vehicles with L2-L5 technology. This could circumvent the car development cycle of manufactures and go straight to the consumer. Once the software is stable and robust the hardware is a matter of installing the sensor package, then mechanicals for gas, break, shift (for automatics, manuals would be a different story). The challenges would come in the form of the liability challenge and certifying a retrofitter to install a system.
by Cybernettr
Unfortunately, around the same time (2029 to be exact) is when Kurzweil predicts strong AI (computers passing the turing test). So it looks like all of these job-crushing technologies are going to come at once, and they’re going to hit us like a ton of bricks. But that’s how exponential growth works — it looks insignificant for a long time and then it suddenly explodes.
by SweetDoug
‘
‘
‘
Can we have a conversation about how my vehicle insurance rates are going to drop to next to nothing when these vehicles arrive?!
Yeah…
No they won’t.
Like an insured driver who doesn’t drive, all you have to do is drive your car once under your own power and that justifies them continuing to charge you out the yin-yan.
That’s why you’ve never heard the Insurance companies talking about this.
Tell me I’m wrong…
•∆•
V-V
by Vladislav
driverless cars in the year 2032 (prediction from 1993):
http://www.youtube.com/watch?v=FlpPR_GgwEc
by Erik
Funny, did you see how thick that iPad was.
by GatorALLin
always funny to look back and see what things they get right and get so wrong…. Much of the older movies that show computer screens and what to display stand out to me. Other than the thickness issue, they did quite well it seems.
by bamford
People mistakenly dwell on death statistics, speed limits, and congestion when discussing auto-drive cars (we will get a name that sticks eventually… “self-driving cars” is the “horseless-carriage” language of today).
The real driver is property damage and medical bills. I was in an accident last month — totally avoidable, three-car collision in stopped freeway traffic. The at-fault driver was “distracted” and cited for careless driving on the scene. My medical bills for that small incident will cost his insurance company is $5K and climbing, plus replacement of one totaled car and a third damaged car. Conservatively this is a $25K incident of stupidity, and this scenario is repeated thousands of times daily, everywhere. Insurance companies will be heavily penalizing human drivers as soon as they can reasonably justify it.
by jmlvu
Consider the effect of having thousands of self driving cars recording everthing drivers do. States could illiminate thier highway patrols and issue tickets though google cars. The tickets could subsidise the purchase of a car making it like the free smart phone.
by melajara
For every tool, there is a user.
As humans couldn’t do otherwise, they first invented dumb tools but for clever users, i.e. human beings or they forced other human beings i.e. slaves to use them for their benefit.
Now that we have the possibility to develop more and more clever tools, an interesting dilemma emerges.
Should we embed intelligence within the tool itself, the easy way, or should we engage a large concerted effort in designing surrogate clever users of the same dumb tools we are using now and from immemorial time?
The first path gives us a Roomba for vacuum cleaning, so called intelligent fridges remembering what they have inside, a paraphernalia of other emerging products, and circa 2030 a car with an embedded driver, rivalling the mobility of a dumb car with a professional human driver, i.e. for each tool we will have a dedicated robot directly embedded within the tool itself and as such having only ad hoc functionality.
The second path would give us Asimo, then sometime ahead, Jetsons “Rosie” and further down Asimov’s Giskard if not Daneel.
I strongly prefer the second path, but by the inherent logic of hyper-consumerism and programmed obsolescence, most surely it is the first path that will prevail, even if some interesting blending is quite possible, e.g. almost universal android robots with an augmented sensorium to ease communication with not so dumb tools or “objects”.
by victorash
…. a machine exploiting another machine :) ?
by melajara
Yes and it makes perfect sense. Why to embed precious resources in a tool sitting idle most of the time when the intelligent machine worker, as for humans, would just have to dispose from an already existing set of tools and, like us (so far), go from tool to tool to perform a set of tasks? This is sheer modularity and traditional division of labor.
Unfortunately it’s at odds with our disgusting society of programmed obsolescence. So, in order to (quite artificially) sustain economic growth, I predict more and more “intelligent” screwdrivers, and so on.
Note that cars are way higher in the toolset hierarchy. As such, it could make sense to have the car and the driver fused in one entity, the autonomous car, if cars really would become autonomous, i.e. perform a collective single task of optimized transportation and traffic, independently of the current notion of car ownership or typical use from and for a dedicated human being.
But even so, will the car do shopping for you? ;-)
by Damon Montano
Sixteen years sounds like a long time.
by GatorALLin
I think the most interesting detail is that car accidents and related deaths could drop to almost Zero with self driving car technologies.
Got me wondering just how many deaths per year we have from car accidents. The first reported motor vehicle deaths for USA is 1899 and reports 26 deaths. That number went up as car totals went up of course. The highest reported year was 1972 with 54,589 deaths in the USA. In 2012 we are still averaging 34,080 deaths per year. Note that deaths represent only a Very small fraction of total accidents and in the USA alone there are over 10,000,000 reported car accidents per year that don’t involve death. http://www.census.gov/compendia/statab/2012/tables/12s1103.pdf
More car accident fatality stats in USA posted here http://en.wikipedia.org/wiki/List_of_motor_vehicle_deaths_in_U.S._by_year
Compare motor vehicle deaths to EVERY death in the USA related to War. The stats for American fatalities and those wounded go back to 1775, or 124 more years that we have records for car accident deaths. So from 1775-2013 there are 848,163 combat related deaths for Americans. Add another 433,161 other related war deaths, then add on all the wounded (1,531,036) and also those counted as missing (38,159), or the Grand Total of 2,756,150. In summary, more Americans have been killed from a motor vehicle since about 1965 than all the Americans killed/wounded/missing since 1775-2013. If you look at only the death rates for all Americans killed in Every war vs. their vehicles you see that 2,434,071 MORE deaths were related to cars and to be fair that includes tracking of Wars for 124 years more than cars even existed. http://en.wikipedia.org/wiki/United_States_military_casualties_of_war
Maybe to help add some perspective of car deaths compared to war deaths, lets look at a recent war of both Iraq and Afghanistan and the totals are from 2001 to 2014 for the USA, UK and all other coalition forces. http://icasualties.org/OEF/Index.aspx The American death toll was 2,302, but there were 22 additional countries that sent coalition soldiers beyond Nato, with UK having many of them at 447 deaths, but even then the grand total is 3,410 in this 13 year period since 2001. (http://icasualties.org/OEF/ByYear.aspx).
Again, to add some perspective of car deaths in America vs. the war in Afghanistan and Iraq for American and Coalition forces, we have more deaths EVERY single MONTH on average in 2007 than in this 13 year war.
We have been averaging about 100 deaths per day, 7 days a week, 365 days a year in America since about 1950, although with improvements in car safety death rates are coming down a bit since 1996, with only 3 years of the last 17 showing an increase in motor vehicle deaths.
One might argue that if you were a government and you really cared about protecting your people, you might at least consider how much money you budget for War and killing people vs. a project like this. I guess the good news is this problem of getting self driving cars to work well is doable and already being done on a small scale. Just down to money, passion and priorities.
by victorash
Thanks for the research , GatorALLin.
Not to downplay it: innocent people die in war (I don’t mean soldiers…), while a car one CHOOSES to have and drive..
by GatorALLin
… to be fair then maybe the logic is choosing to fight in a war, vs. choosing to drive to work seems like the end result of death is much more extreme… Surely innocent people get run over or killed by drivers every day.. of course war kills many innocent people to damage with wounded or lingering emotional/mental issues well beyond just the combat parts when they return home (drug abuse/alcoholism/self medication/etc) so it may be unfair to count the damage of war by just death/wounded. War is a very complicated problem of course… at least the idea of self driving cars seems like a problem they have solved and using logic of More’s Law, it seems only a matter of time/passion to get cost down and usage up. Where government could step in to advance the speed of certain developments is to allow discounts with taxes or laws or other advantages to dramatically speed this up (especially in the case where thousands of lives per month could be saved in each country). Of course we can also look at issues like curing cancer, etc… and of course there is millions of fund raisers to save lives for breast cancer or a hundred other great causes. Just no “fun runs” or March of Dimes drives for self driving cars. Of course also if the government had to hold a bake sale to raise money for their next nuclear sub or next drone advancement there would be a lot less money spent there (or some billion dollar brownies).
by GatorALLin
** Moore’s Law **
by beephatron
You could prolly save 90% of lives making helmets, roll cages, and other safety equipment mandatory. A lot easier things we can do than 30 year down the road tech. Obviously people value their stupidity. Do you wear a helmet while driving?
by Editor
Please see UPDATE at the bottom of the article for a brief interview with the principal analyst of the IHS Automotive autonomous cars study.
by smb12321
We are fast approaching the Impersonal Auto Age where one simply calls for a pickup, pays a low fee and arranges a return trip. Just as public transportation is cheaper than a car in large cities, this system is far cheaper than paying for a car, insurance, taxes, maintenance and end up with a loser investment.
This introduces another problem. The federal gov’t survives only via heavy debt, fees at all levels and double / triple taxation. Technology reduces price and thus taxes. What happens when fund sources (cars, gas, consumer goodsl etc) disappear?
by someday69
pessimistic,Because,,What if the’people of = then name the most -Kewl’est’Country..an’Yes’well”’They all have ‘Safe’driving’,an’Knowone’is getting hurt’no’speeding tickets,no’crash’n'burn’.NO’more!..We want,it.NoW! TOO!!
The country that can [print its own money} Will just absorb’the cost’of SMART ,,ROADS..an’ smart cars..= no problem’}
by granitet
These forecasts do not take into account the exponential nature of our growing knowledge and rapid adoption of new technology. We will be meeting their estimates much earlier than the current projections.
by Editor
Can you provide any evidence to support that in the area of self-driving cars?
by JoseLuis Malo
Well, I’ll try.
Nissan says quote
“We’re going to get there even sooner than we think,” Ghosn said. “What’s going to be left is the reliability of the system. … 2020 in my opinion is going to be the latest because we’re going to be under pressure from a lot of competition. The pressure is now on us [Nissan] to be sure we are bringing the first cars on the market.”
Self-driving cars will hit the streets by 2020 says Nissan at CEATEC
http://www.extremetech.com/extreme/167897-for-sure-self-driving-cars-by-2020
And Volvo also reiterates, quote
“Autonomous vehicles are an integrated part of Volvo Cars’ as well as the Swedish government’s vision of zero traffic fatalities. This public pilot represents an important step towards this goal,”
Volvo to Test Self-Driving Cars on Public Roads in 2017
http://www.dailytech.com/Volvo+to+Test+SelfDriving+Cars+on+Public+Roads+in+2017+/article33859.htm
Volvo sees crash-free car by 2020
http://reviews.cnet.com/8301-13746_7-57594979-48/volvo-sees-crash-free-car-by-2020/
by Editor
Jose,
Thanks for the info. The problem here is the vague, imprecise language used by marketing people. Marketing statements are not a useful source of accurate and precise information (to put it kindly). ExtremeTech says: “Automakers who say they will have self-driving cars by 2020 typically mean the middle of three levels of self-driving, or the highly automated car level.” That’s too vague to comment on. Same problem with Volvo.
The IHS Automotive study is based on an in-depth analysis of technical capabilities, with five precisely defined levels (see diagram in article). The IHS executive summary of the study shows Nissan reaching L3 by 2020 and Volvo reaching L2 by 2014+, with no evidence for reaching L3. None of the vendors have shown evidence of reaching L4 by 2025 (data beyond that is not shown), except for Google (2017+), but it appears that the Google development still needs to be converted by a car manufacturer into a deliverable, tested car.
Having followed Juliussen’s excellent analysis work for a long time, starting with his work in computers, I’m confident that his data is more accurate and credible than vendor marketing data.
by JoseLuis Malo
Well you have a point as far as marketing and evidence, and I don’t know if I’m making wrong assumptions but given the fact that Kurzweil predicts a 2029, a mere 15 years from now, human level AI (I’m assuming a normal average adult) with emotions and everything, why would it take all the way to 2035 to get just a 9% of the total amount of vehicles on the road, this being a much smaller technical problem compared to the human level AI?
If the 2029 year is correct, by 2035 we should have people like AI as ubiquitous in society as today’s laptops given the additional facts that nanobots are also supposed to be everywhere including in most people’s bodies and brains. Autonomous AI drivers should be ubiquitous way before 2035.
by Editor
Yes, this is an important question. We haven’t looked at the impacts of intelligent machines on self-driving cars yet or, perhaps more important, how humans will mesh with such machines in this case. In general, the impacts will depend on which processes in the automobile industry involve information, and intelligent machines should significantly accelerate development for those processes. For the current news article, we’re focusing on the IHS Automation study.
by whitepheonix
Pls excuse me miss editor if I sound rude…but did u read “Singularity is near” or any of Ray’s speeches?This is kurzweilai,right?(regarding granitet’s comment)
by Editor
Yes, also headed research and editing for The Singularity Is Near, so I’m fairly familiar with it. :)
by whitepheonix
Ray Kurzweil shows and proves in his books,films,speeches that every field that is an information technology(self driving cars in this case)follows an exponential growth . Thus Granitet’s comment is proved accurate by Ray Kurzweil himself,for me at least …
by Editor
For us, it’s a given that forecasts that don’t take accelerating intelligence into account may be inaccurate. Determining the extent of the inaccuracy for each case is a non-trivial exercise.
by smb12321
I thought exactly the same thing. I mean, there are functioning safe cars today that have driven over a million miles without an accident. So why are we waiting a quarter of a century to “have” them? Something doesn’t add up.
It’s like saying we have wi-fi but will manufacture products to use it in 20 years.
by socean
This report is designed to calm automotive industry related enterprises, to make them feel they are in control of the rollout of autonomous vehicles, and to assuage fears of disruption and job losses. Its also designed to give policy makers excuses for delaying the sweeping changes that will need to be made.
We have self driving cars now, today, not sixteen years hence. They work great, they’re just too expensive for personal use. However, Its TRUCKING and commercial vehicles that will be automated first. And even at current prices, its possible to recoup the investment.
This report is just more “sanitized for CEOs protection” B.S. The robot vehicle revolution will not wait. Its here now, and its just going to gather momentum.
by Editor
Do you have any evidence to support your claims, or are these just speculations?
by Erik
About 10 million self driving cars to be sold in 2035.
They are too pessimistic, six years ago nobody had a smartphone. Now more than 1 billion units have been sold…
by dougw659
Ummmmmm…Smartphones essentially cost nothing when you sign up for a plan that was maybe $15/month more than your existing cellular plan….Auto-Drive Cars will cost upwards of $35K…..maybe not quite the same adoption rate for such a major price difference. Also, if your Smartphone didn’t work correctly you might miss a call or not be able to browse the web (…jeepers!). If your Auto-Drive Car doesn’t work correctly you might plunge off the side of a ravine or crash into a fully-laden petroleum carrying truck on the interstate…again a minor difference that the culture may not find so easy to absorb….
by Erik
I’m not saying 1 billion cars will be sold by 2020, but the price difference for self vs manual drive will be small compared to what the rest of the car cost. Most likely it will be cheaper to have a self drive since insurance costs will be lower By 2035 people will have changed cars many times.
Few people in China, India and Africa have a car today, it will change in 20 years.. + all the elderly, handicapped etc. that can’t drive.
I think it will be the other way around, people will be afraid to travel in a car that doesn’t have a self drive system. Most people already think a car with ABS is safer than one without.
by Gear Mentation
What I think is pessimistic is that the price will not go up that much. Maybe in the beginning, but for fully autonomous cars, you also save some on driver controls. This is especially true because there just isn’t a general market for more expensive vehicles.
by Editor
Can you support your generalization from relatively simple reductions in chip size and improvements in performance (in the case of cell phones) to highly complex car-driving technology that will require fundamental technology breakthroughs in different areas?
by Gear Mentation
Are you replying to me? Hard to tell with this format. Anyway, I can’t support with specifics- but aren’t you quoting predictions from people who don’t know about the Singularity? So the price comes down hugely, plus I don’t believe (barring a basic minimum income or unionization etc.) that there could possibly be a market for a car sold to the general public for thousands more. Sorry I can’t give more…
by Editor
Gear Mentation: I wasn’t replying to you, but thanks for the comment. Not sure why knowing about the Singularity is relevant or why you think Dr. Juliussen is unaware of it. We’re discussing relatively near-term forecasts in consumer-related automotive technology, for which he is an eminent authority and also an eminent expert in futures studies, including technology acceleration trends and pricing models.
by smb12321
Philip Tetlock’s massive study of experts and their predictions revealed how utterly wrong they were. Most predictions were based on ideology or rooted in their own domain without outside references. I have increasingly grown wary of experts (particularly in academia) who crow over a 1/20 success ratio. After all, if the experts were rights, no one would ever lose in the market, hurricanes would be bad and frequent (NOAA), computers would not be widely used (IBM), the USSR would still be around (CIA) and nuclear energy would not exist (Einstein).
by Editor
Yes, predictions based on biased personal judgement and speculation, especially in areas of high noise (stock market, etc.) are usually of low value, as in Tetlock’s study: “Expert Political Judgment: How Good Is It? How Can We Know? (2005) describes a twenty-year study in which 284 experts in many fields, including government officials, professors, journalists, and other, and with many opinions, from Marxists to free-marketeers, were asked to make 28,000 predictions[1][2] about the future, finding that they were only slightly more accurate than chance, and worse than basic computer algorithms.” — http://en.wikipedia.org/wiki/Philip_E._Tetlock. IHS Automotive (http://www.ihs.com/industry/automotive/oem.aspx), a division of IHS (http://www.ihs.com/index.aspx), does large-scale multi-client research projects covering more than 130 auto makers across 90 countries and stays in business by being reasonably accurate. The current study involved in-depth analysis of 19 automobile and associated companies involved in self-driving cars.
by JoseLuis Malo
Although IHS 1 to 2 years forecasting could be impressive, how is IHS track record for 10 and 20 years predictions?
by Editor
Good question. I don’t have any information on that.
by Gear Mentation
Yes, did you ask him if he thought people in general would buy cars for thousands more? When young people are in the process of divesting themselves of cars? I thought he was probably unaware of the Singularity because most people are. They don’t take it into account with their predictions.
by Editor
Good question. The IHS Automotive executive summary states:
“Self-driving ‘only’ [level 5, fully automated, no driver required] cars … will have lower prices, as no driver controls are needed. The SDC electronics price premium will start at $7K to $10K in 2025 and will decline to over $5K in 2030 and over $3K in 2035. Total SDC sales are expected to grow to $11.8Min 2035. The long-term opportunities are very large as nearly all autos in use are likely to become self-driving cars or self-driving commercial vehicles after 2050.”
by Erik
The reason we couldn’t build a self driving car 20 years ago was because we didn’t have the computational power to do image analysis and 3D reconstruction in real time (360 degrees, 100 times a second).
It’s the same problem we had 35 years ago, when people still were using type writers.The computers at that time didn’t have capacity to hold the text and later graphics in memory, at least not at an affordable price.. Moore’s law changed that and 15 years later nobody bought a type writer.
Same thing will happen with cars, and it not going to take 35 years for us to get there, more like ten years once the first self drive has been sold.
by girsa56
what ever happened to 2020?
by Editor
Which forecast are you referring to? Some “self-driving” forecasts are really for L4 (with human) — see diagram.
by Cybernettr
In The Age of Spiritual Machines.” Ray Kurzweil predicted for the decade ending with 2019 (this decade) that “Automated driving systems are now installed in most roads.” He didn’t say they would be common, just available. However, he also didn’t say whether they would be fully or only partially autonomous, so I guess these predictions are not in conflict.