Driverless vehicles, on their way

Uber passengers in Pittsburgh will be able to summon rides in self-driving cars with the touch of a smartphone button in the next several weeks. Uber also announced that it is acquiring a self-driving startup called Otto, co-founded by Israeli Lior Ron, that has developed technology allowing big rigs to drive themselves.

Via Mark Thorson, here is more.  And in Finland:

Residents of Helsinki, Finland will soon be used to the sight of buses with no drivers roaming the city streets. One of the world’s first autonomous bus pilot programs has begun in the Hernesaari district, and will run through mid-September.

Finnish law does not require vehicles on the road to have a driver, making it the perfect place to get permission to test the Easymile EZ-10 electric mini-buses.

So perhaps Finland can become a market leader in this area.


Always check first Switzerland

Much depends on the actions of the car companies, GM and Ford in the US. Of course, they have every reason to slow down the pace of development: even if "cars" survive the transition to autonomous vehicles, there will be lots fewer of them, not a great prospect for "car" companies. And as I have commented before, "self driving cars" is a euphemism for "self driving transit": millions of autonomous "cars" racing about at 70 mph is absurd, more like demolition derby than Indianapolis 500. What the car companies prefer is to preserve (sales of) "cars" as we know them and develop autonomous vehicles that would supplement not replace "cars". That means one of two things: "self driving cars" that have maximum speeds of about 25 mph (what Google envisions) or "self driving transit" that is an alternative and requires a separate right of way. In the meantime, people can dream of flying cars in every garage and spaceships to Mars for the summer's family adventure.

Wasn't it on this very blog that it was predicted self driving cars would lead to far more miles driven?

Uber envisions a future when few people own "cars", instead relying on self driving Uber cars for transportation, meaning that there will be lots fewer "cars" that would be driven many, many more miles. That's a future I find appealing but the car companies most definitely don't.

These Uber "driverless" cars all have "human backup drivers".

So what is the correct, truthful term for a driverless-car-with-a-driver ??

The B.S. media hype on driverless cars just never ends.
Genuine driverless car capability is decades away.

If we were honest we'd call this advanced cruise control. People are used to cruise control, using it with caution, watchodogging it.

@anon: does your cruise control make turns for you and know where to stop to let you out? Because these cabs will.

"These Uber “driverless” cars all have “human backup drivers”."

That will almost certainly last less than a decade. Humans don't perform well as "backup drivers" (i.e., taking control in an emergency).

Further, autonomous cars will improve incredibly rapidly. Every single accident will provide data on how to avoid that type of accident in the future. Contrast that to human drivers, where over 10,000 people die every year from drunk drivers...and it just keeps happening, year after year.

The smart way to test really-automatic-cars, assuming you want to get really-automatic-cars to work, is to have a professional sitting there in case of emergency.

Ya, people on the Internet will make fun of you. Big whoop. In a few years, assuming the tech works, you can get rid of the professionals and then you have your really-automatic-cars.

"Wasn’t it on this very blog that it was predicted self driving cars would lead to far more miles driven?"

"Of course, they have every reason to slow down the pace of development:..."

Doesn't matter -- it's not a monopoly industry (nor confined to a single country). Even if the success of driverless vehicles was a certainty and this was certain to result in fewer vehicle sales, each automaker would still need to pursue the technology vigorously to ensure it was not shut out of the new market.

That said, I think predictions that driverless vehicle would kill the market for individual vehicle ownership are greatly overblown -- pushed by Uber (for obvious reasons) and seconded by pundits who don't quite realize how unrepresentative they are in living in dense urban areas where congestion, long-commutes, and scarce parking are major issues. Elsewhere in the U.S. (where the vast majority of Americans live and work), there will still be major convenience/status/comfort advantages to owning a personal vehicle (whether it is driverless or not). Most people don't buy undifferentiated transportation appliances, they buy vehicles to suit their self-image (as they do clothing, homes and furnishings). And people do things with their vehicles that don't fit the Uber, just-in-time pickup and drop-off model. They go away for the weekend. They tow trailers or haul kayaks to the river or mountain bikes to the trail-head. They pick up loads of building materials at Home Depot, etc.

And keep in mind only a small percentage of Americans use Uber. Only 15% have *ever* used a ride-hailing service and 3/4 of those use the services once a month or less, which leaves just 3% of Americans who use ride-sharing services regularly:


For cities self driving cars shaped like those at the link would be great:

Some great points, especially about people buying vehicles to suit their self-image. But....

"They pick up loads of building materials at Home Depot"

In a "self-driving cars" utopia, the materials you buy from Home Depot are delivered via self-driving car.

(As I understand it, the real potential of self-driving car technology won't be replacing human drivers of the car. It will change what you think of as a car. Order a arrives in a little bot. That kind of thing.)

" Even if the success of driverless vehicles was a certainty and this was certain to result in fewer vehicle sales, each automaker would still need to pursue the technology vigorously to ensure it was not shut out of the new market."


"That said, I think predictions that driverless vehicle would kill the market for individual vehicle ownership are greatly overblown"

I agree with this, but with caveats. I expect to see a reduction in the vehicles owned by American families in the suburbs and rural areas. So, families might go from 2 or 3 cars down to 1 or 2 cars. Though, it might be less than even a full car. Still I would expect a downward pressure, as they'll be less need to keep and extra car for the occasional trip.

Isn't car lifetime in non-northern climes primarily a function of kilometres driven and not age of the car? Fewer accidents will reduce the number of cars to be manufactured, but that would not reduce the number of cars on the road.

We'll see. Someone asked why with an early start and tons of research, Google wasn't first to market?

I said maybe with an early start and tons of research, they know better than anyone that this is a bad idea.

As indicated in my first comment, Google envisions autonomous vehicles with maximum speeds of 25 mph. Would there be much demand for that? And what one must remember about Google, it is an advertising company (notwithstanding all the hoopla about "tech"), the appeal to Google of a self driving car being that it can take the customer the last leg to the businesses that advertise with Google.

Google invented massively parallel computing, and a raft of massively parallel applications, including various AIs, such as they are in 2016. For a while there was some question how they were going to make that pay, giving most of it away, but yes, advertising was the answer.

Now when an acknowledged leader in AI only trusts 2016 tech at 25 mph, you in your wisdom call them wrong?

You do understand that one of the reasons that the Google vehicles are limited to 25 MPH is so that they are classified as NEV's right?

Sure, and not coincidentally the 25 mph limit for NEVs is right at a safety boundary. Much less bad things happen below 25 mph.

To further give you a clue here, it is possible that Google used self driving cars as a vehicle for AI research, and they found the current limits. If the limits are too low for deployment, you do not sell a car, you go back to the lab on AI research.

I absolutely agree with Google: self driving "cars" will likely be limited to 25mph. What I have presented are alternatives: self driving cars at 25 mph (the Google model), self driving transit that has a separate right of way and much higher speeds (that supplements but doesn't replace cars as we know them today), or Uber self driving cars (that replaces individual ownership so there are lots fewer "cars" traveling many, many more miles at somewhat higher speeds than in the Google model).

I seem to recall Cowen predicting the self driving transit model, although he didn't call it that. Each model has its advantages and disadvantages, and each model has its supporters and detractors, the car companies really not liking the Uber model (because it would mean far fewer sales of "cars"). Are the car companies powerful enough to determine the model implemented?

I guess if I wanted to be out there I'd note that City Bikes are shredding self-driving cars on ridership, safety, economy, and health.

Perhaps there are several opportunity costs associated with People Movers.

Anon hits it on the head by mentioning bicycles and people movers: the transportation chosen should be based on time and distance. Instead, what we have is essentially two means of moving people, cars and planes, the former used anywhere from a trip to the grocery store around the block to the vacation in Florida. It's the dominance of cars for a hundred years that has shaped the design of cities even though cars are highly inefficient (but convenient); and not surprising, the cities are highly inefficient. If self driving "cars" can get people over their infatuation with cars, I don't care if they believe (falsely) in a model that will never exist (i.e., autonomous "cars" racing about at 70 mph). Walking, people movers (like in the airport), bicycling, self driving cares, self driving transit (including "cars", buses, and trans), and airplanes. A revolution in travel that would catapult the economy into the 21st century. We are on the cusp of developing the technology, but do we have the will. With so many interests at stake (the car companies, the road building industry, the oil and gas industry, etc.), I'm skeptical. Hopeful but skeptical.

Perhaps Tyler will ride some bicycles in Denmark and come back a New Man.

Confusing Danes with Dutch, anon

As I've acknowledged, the Dutch are number one per capita bicyclists, but Danes are number two.

What's with all this fixation with travel, anyway? With the growth of the orthographic symbolist workplace and the decline of the object manipulators, physical location should be less and less important in the economy. It's becoming unnecessary to travel for food, clothing and other consumer desires which are now increasingly being delivered to the buyer's door. A growing percentage of travel is of an optional, discretionary nature, a trip to the museum, a vacation in the Bahamas, a Cubs game at Wrigley. The travel patterns of the average human, American human, at least, are likely to be much different in the not too distant future.

Wow, I'm so impressed that City Bikes is "shredding" a technology that has not been developed or deployed yet, that's so great for them.

No. Because they wouldn't stil be investing in it if that were the case. My guess is that for a company that large, in so many product categories, that the importance of a good rollout is much higher than the importance of getting to market first. If they roll out too early and there's bad press, considering expectations, the whole company takes a hit. With the other manufacturers, I think people are quite used to car recalls for safety reasons after a number of deaths, etc., and generally benefit from lower expectations which mean that the cost of screwing up is lower. Which might make them more interested in prioritizing being first to market.

Putting out the first press release about your driverless car (*) does seem to be the way to be considered a leader in this field, yes.

(*) Driver required.


The Pittsburgh experiment will include a driver behind the wheel of the "autonomous" car and an observer in the passenger seat. So much for vehicular autonomy as it would be understood by a normal person.

And the driverless buses in Finland will not be "roaming the city streets".

Those French EZ-10 buses are restricted to pre-planned routes that the bus has "manually" driven before to learn the terrain and environment. There are onboard sensors to detect "normal" obstacles and pedestrians. Sensors can not handle the Finnish winter weather. Average operational speed of these buses is 6 MPH.

This Helsinki bus operation is a brief summertime test of the system, with safety observers. There's no conventional steering wheel on the EZ-10 bus, but special test equipment permits alternative control of the vehicles for test and safety purposes.

It's so outrageous that somebody would want to test these driverless AI vehicles before fully deploying them! Are they insane??

Does this mean that the chances for a collision with a vehicle operated by a human driver yakking on a wireless phone or texting on some device will be higher or lower than the chances for a collision with a vehicle operated with no human driver, whether in Pittsburgh or in Finland?

Put a self driving lane on I-5 in California - 100 mph allowed and required, autonomous vehicles only, ridesharing encouraged. Or, we could go with a $100 billion high speed train. Door to door, the self driving lane would always win.

+1 for high speed trains as not necessarily the most prudent investment in 2016.

You're not factoring in all the costs. Trains being more energy efficient and all ...

Commuter trains aren't significantly more energy efficient. They make too many starts and stops. Freight trains are where energy efficiencies dominate.

Thanks. I hadn't thought of that. I guess it also depends on ridership.

Personally, I like the notion of high volume routes being outfitted so that the normal wheels of the car are extended somehow to ride on rail, as a sort of hybrid road/rail personal vehicle. So instead of the normal onramp, you get to the onramp, hop on the tracks, and away you go with rail efficiency. A number of obvious technological or systems design issues come up, which would be a little tricky, but are easily within the realm of possibility in the near future if it were considered as of interest.

But ... path dependence. I doubt anything like that will happen.

Tyler, you got cybergoobpunked. Uber will not be introducing driverless taxis anytime soon.

Same with 3-D printers. I thought the posts on that would never end. When it turns out your wife would rather buy patterned porcelain than the dirty-yellow resin plates you're pumping out in the garage, the 3-D printer revolution sputters to a halt.

It seems like the most interesting applications have been medical. In other uses, scale economies tend to apply, which sort of defines that it is only useful for small markets and prototypes, etc.

But internal IP involved in prototypes probably doesn't make the news any more because it's not new and cool, so we don't hear much about that any more.

One benefit (the only benefit) of getting older is that you can recognize bullshit because you've seen so much of it. Glenn Turner and Dare to be Great? Check. The two largest (digital) advertising companies (Google and Facebook) sold as "tech? Check. Two skimmers (the credit card company and Paypal) rather than one skimmer (the credit card company) sold as an improvement? Check. Self driving cars and spaceships to Mars? Check. People will believe anything.

I would not call myself an expert, but I am a retired programmer who has done the Stanford Machine Learning MOOC. Based on that domain expertise I would certainly say Google is near the top in AI. Details and confirmation here:

The "advertising" thing started with Apple fanboys, and people who don't know, just don't know.

rayward has been beating this incorrect drum for months. No word about what the ads we all ignore pay for: free email, search, Youtube, Docs (and a million other services) for Google, networking globally with billions instantly (and a million other services) for Facebook. But sure it's not tech. Nothing new there. I know this because rayward keeps typing it. Very prior_approval.

"Merrill lynch now believes that a stand-alone valuation potential for YouTube may be as much as $90 billion" -

Youtube isn't a freebie, it's an investment.

Way to miss the point, Nathan W. Youtube is free to watch for the user, ad supported exactly like 'free' TV. But according to rayward, there's no technological difference between Youtube and TV, nor TV and radio, LOL. He's just laughably wrong on this particular obsession of his. His inequality schtick is less stupid.

I agree that he's wrong. I'm just saying that YouTube doesn't subsidize free email, etc. It is profitable on a standalone basis, to the extent that estimates of NPV of future profit flows are to the tune of $90 billion, according to that estimate. That's not a loss leader, that's a profit centre in the making.

Don't forget 3D Printing.

We never read about that here anymore!

It's hilarious how everyone is sooo invested in pooh-poohing driverless cars. There's nothing the Internet hates more.

This is true. Driverless cars and women.

The obvious application is driverless buses with no more generously pensioned drivers. But that still doesn't solve the How-Cognitive-Elites-Can-Avoid-Rubbing-Shoulders-With-People-Not-Like-Them problem.

So, I've heard that Pittsburgh is becoming sort sort of robotics tech hub, which is leadind a revival and gentrification of the city.
Can anyone lend credence (or not) to these rumors?

I know one very well-paid individual who works for an IBM vendor, so yes there is some tech presence there.

To add (fire) to the conversation, here's a screenplay making the rounds about a self-driving car disaster and the ethics/implications of commercial automation:

The idea kind of reminds me of Edison electrocuting animals. He was correct that DC is safer than AC, but he ignored the magnitude of the risk. AC might have been more dangerous, but it was still far safer than a lot of dangers that people routinely face.

Self-driving cars will have their share of accidents. But if they're a better driver than your average human, then the shear convenience they offer will ensure adoption.

"Self-driving cars will have their share of accidents. But if they’re a better driver than your average human, then the shear convenience they offer will ensure adoption."

Self-driving cars raise the possibility of the entire fleet improving from the results of one accident. The fleet of human drivers doesn't have anything close to that ability.

Comments for this post are closed