Driverless Cars Are Taking Longer Than We Expected. Here’s Why.

Is there a chance that we are addressing this problem backwards? What if only driverless cars were permitted on a roadway where other vehicles and pedestrians were forbidden?It would be relatively easy to keep them properly separated Even in adverse driving conditions.

The tech companies are under the flawed assumption that a vehicle is just a mobile phone on wheels that happens to transport people. Unfortunately it isn’t. Tesla is one of these tech companies btw. Just wait for it....

I have trouble understanding what problem autonomous vehicles are supposed to solve (except perhaps bored software engineers!). The promises of a vehicle that can operate safely in all road conditions assume a perfection of mechanical and electronic systems which has never been achieved. For example: our national security agencies are rightly terrified of what someone hacking into a major national computer network might do. It would no doubt be far easier to disrupt a network of cars, let alone individual cars. But simple system failure will be probably be more common, but could be equally disastrous. Electric cars respond to the greenhouse gas emissions and air pollution that fossil fueled cars produce. It's a misnomer to say that electric cars are "zero emission" as there will be some emissions in the manufacture, maintenance and disposal of the vehicles. There will also be emissions in the larger amount of electricity generation which will be needed. But overall electric cars represent a feasible solutionto a real problem. As commenters have pointed out, autonomous vehicle proponents seem to imagine a world without pedestrians or transit riders. The MIT Media Lab ran an autonomous vehicle simulation without any pedestrians. Perhaps the "problem" that autonomous vehicles solve is maintaining a nation of suburban sprawl as the fossil fuels that built it are becoming scarce. Then Americans won't have to face the terrifying prospect of walking or taking the bus.

I would respectfully disagree with Dr. Shladover. When something goes wrong at 30,000 feet, you drop 30,000 feet to the ground and a couple hundred people are killed. When something goes wrong on the road, the vehicle can simply pull over, or at worst, come to a stop in its lane. That's why we let random people drive cars, and have two extensively trained pilots on every passenger aircraft.

This is not a very good article. Driverless cars are restricted by weather and mapless areas. That’s it’s only real piece of information. The rest is more of less fluff.

Why does no one seem to be picking the low hanging fruit here? Open road interstate driving. Straight, no turn, point to point buses. People rode in cars for years before there was automatic, power steering, air conditioning. But it seems like driverless vehicles have to do it all out the gate.

Driverless cars (where I DO NOT have be ready to take control) are 20 years, many injuries & many deaths away. Why? You can't test the car in a simulated real-life test environment. You have to do it on the open road and designers can't anticipate the permutations & combinations of events on the open road. So when the car encounters an event or series of events it hasn't been programmed for, it selects an inappropriate action or 'freezes'. Don't believe me? How often does component of your home network require a reboot: turn off its power, wait 30 seconds, power it up? Or the: airline systems, financial systems, check-out systems, retailer's website that is overwhelmed? In each case, it's a system failure that does not result in your being injured or killing you - so we live with it. Another scenario is the car 'knowing' the local traffic laws. Here are the laws for the right shoulder on a highway: -Florida: When you see a First Responder, either move to a left lane or slow to 35mph and pass -Massachusetts: On certain highways at certain times on certain days, the right shoulder is a travel lane (you can do 65mph) The car will have to 'know' where it is, 'know' the applicable laws and then react correctly. Or that puddle on the highway after a heavy rain. Go to the other lane to avoid or plow right through and risk losing control? Watching the car companies trying to get this right will be entertaining for years - just be sure to avoid getting hurt or killed.

After having driven more than a million miles in four countries, I look forward to the day when driverless cars are as common as driven cars. I'm newly retired and I hope that my next car will be driverless, or what is retirement for? As we get older, what a terrific private ambulette the things would make for the various and ever more frequent doctors visits old people make, not to mention the occasional minor emergency? They would enhance our independence rather than take it away as the loss of driving privileges for senior citizens many jurisdictions are contemplating now tend to cause. I could probably pay the cost of such a vehicle by just not having to take so many taxis.

As a 75 year old, I and probably all seniors, are very not looking forward to the day we can no longer drive. Self driving cars can’t happen fast enough.

In a culture that follows the rules, this level of automation may be possible because the background situation would reduce risks and increase predictability. Does that sound remotely like Americans?

My late-model sedan has many of the latest innovations (adaptive cruise control, lane departure warning, blind spot detection). All are flawed, if not dangerous in and of themselves. The adaptive cruise control has hit the brakes on a empty highway when it mistakes a bridge abutment for another vehicle. I've disabled the lane departure warning, because I regularly have to move over for cyclists on suburban streets, and got tired of its nagging. A feature that reads speed limit signs and posts them on the instrument display is often wrong, including mistaking '70' for '40' mph, and even displaying 100 mph (why would the programmers even think this is a valid possibility for a car sold in the USA?). While not self-driving, the regular failure of these so-called safety features indicates sloppy engineering and a lack of real-world understanding. Maybe the more talented engineers were assigned to the self-driving program?

If you read the final report on the death mentioned in this article, you’ll see that the neural network hadn’t been trained on images of a woman pushing a bicycle. The software cycled through choices and finally froze. The driver wasn’t paying attention. Expect frequent updates of software via the internet and endless discussion of where liability lies. I question whether the mapping mentioned will ever be complete or perfect. Not looking forward to this technology. Universal enforcement of no cell phone use via software would go further in reducing accidents.

It's strange how the author doesn't talk about Tesla. Their cars drive well enough that "pictures of sleeping Tesla drivers" is a thing online.

It's going to be a decade at the very least, and I expect much longer. Oh these things will be able to drive on a well-maintained freeways in decent weather, sure. But odd city and country roads, in bad weather? There are so many micro decisions that a human makes while driving I think machines will be stymied for a long time to come. Many of these decisions are not reducible to yes or no -- ones and zeros, that is. A dog runs out on a foggy night. Or is it a child? There are people on the sidewalk. Do you swerve if there's not enough space to stop? Maybe try to seer into a tree to stop the car. But if it's a dog you may have to hit it to save humans...complicated stuff. Then add computer glitches, which we all see regularly. Nah, it'll be some time.

My serious prediction: in 100 years the present fascination with autonomous vehicles will be studied as a perfect example of scientific fad and popular delusion. It isn't ever going to work, as currently conceived. Those of us who say that are routinely discounted as old fashioned; but in fact it's that the enthusiasts don't want to hear about problems; reinforce each other, and just keep their hands over their ears. The problem: mechanisms to provide "autonomy" - are just that; mechanisms, designed and built by - humans. The fact that they are designed by committee and astoundingly complex doesn't help. All "mechanisms" can, and eventually will, break or malfunction. Age; crazy ants in the electronics; water leak; cosmic ray hits, bribable inspectors- the list of factors capable of causing mechanical and or electronic mechanisms to fail is very long. It doesn't matter how "nearly perfect" they can be made- at some point, an autonomous vehicle will malfunction, in a very bad place and time, and will cause tremendous damage and loss of life. Will, not might. The public will not like that when it happens. Restrictions will then be put in place, making everything more complex and less efficient, until any benefits disappear altogether. Enthusiasm will fade sharply. Profits will never materialize. And this is all 100% foreseeable. Car makers, realize: all accidents will now be - your fault, not the driver's. Is that really what you want?

For autonomous to work they need to be networked and talking to each other. They should also only be 10 passenger van-like ride shares.

If we would only spend the time on public transportation that we wasted on AV’s we would be further ahead. This will be difficult in a technical illiterate nation like the US where we cannot even build a train or train network.

"Who is we in ",,, longer than we expected"? Don't expect widespread sales of driverless cars for another 10 years.

Unsafe at any speed.

The reality is that autonomous cars won't become a reality absent a dedicated right-of-way for them. Of course, a vehicle on a dedicated right-of-way is . . . .transit. Duh. The problem with the Boeing 737 Max is a defective design: defective software wouldn't be needed absent the defectively designed aircraft. Indeed, Boeing knows the aircraft has a defective design: why else would Boeing install software intended to offset the defective design. Duh. Why do journalists write articles that are essentially myth? Is it because they believe readers can't handle the truth?

Would I ride in a driverless car? (Seems we're being told what to comment on now.) If their safety is proven, if I have some control in cases of emergency - or changing my mind about where I'm going, if it's comfortable, if I can afford it. But the problem with driverless cars is going to drivered cars. In order for driverless cars to be safe, they have to be the only cars on the road. As long as even a few human-guided cars are on the road, driverless cars will be vulnerable. And those holdouts aren't likely to be the most courteous of drivers.

When will autonomous vehicles be common in the US and elsewhere? How many positions for cabdrivers and large-truck drivers will there be in the next several decades? See the new projections, with reasons behind them, at http://worksnewage.blogspot.com/2019/06/driverless-vehicles-and-driving-jobs.html.

If all cars on the road were AV capable and the infrastructure of those roads were fitted with supportive technology, the reality of Personal Rapid Transit would be here. The biggest obstacle is public infrastructure including real time data on roads - condition, construction, sensors, etc. It took a long time to turn roads fit for horses into auto friendly thoroughfares. Hopefully we'll be more pro-active this time.

The idea of autonomous cars is usually connected with ride-hailing services. I would like to purchase my own autonomous car. I see the driverless system as being similar to cruise control - I turn it on in the proper conditions, and I take over driving in places where I am the better option. This is much closer to fruition than 100% driverless ride-hailing vehicles that can operate anywhere. I would engage self-driving on the freeway or suburban streets, and take over driving myself when it is snowing or when I'm driving through a parking lot looking for a parking spot near the store I want to visit. Lat week as I left a Cleveland Indians game with hundreds of people crossing the street and a police officer standing in the intersection directing traffic, waving cars to go when the light was red, I thought "How will an Uber ever handle this situation on its own?"

When driverless cars malfunction, who becomes responsible for accidents? Once the driver is removed from decision making, it becomes the manufacturer’s responsibility; they should be pulled into the insurance settlement negotiations and certainly into any lawsuits.

I have not seen any reports of "fully autonomous" vehicles being tested during winter or bad weather. You might as well call these vehicles "sunshine cars" because it appears that the only place they can be used safely (?) is in areas with no winters.

I really don’t think so. My new Ford 1-Ton F-350 is my dream vehicle for what I do. It’s interesting that at times I can’t quickly understand the various beepers going off - lane control, passing vehicles, etc. already too distracting. I like to drive and control things, giving that up would be very difficult. Actually impossible. I don’t trust my I Phone 100 percent since there are always hiccups.

We are importing millions of low skilled immigrants into our country at the same time we are developing technologies which make their labor less valuable. We have the technology to harvest crops with Artificially Intelligent Robotic harvesters. We currently have the technology to put driverless long distance trucks on the highways. Staffing in stores will be reduced further with AI security and self check-out. There are millions of internet connected cameras recording our streets. Privacy will cease to exist. There will no way to protect your communication or interactions when every phone call or trip is digitally recorded. Both Democratic and Republican politicians seem to incapable of even discussing the issues of population growth and privacy.

Sure, I would ride in a driverless car in Disneyland set up or even Phoenix taxi - but in Mumbai, Jakarta or even Bogota - no way. In my opinion, this whole concept of driverless cars is folks having solutions go looking for problems to solve. A driverless car can only be driven where maps are drawn to inches. How could one suddenly change his mind and head for a McDonalds rather than a Wendy's - slightly out of the way. I am an engineer with a Caltech degree and 43 years of experience in the oil patch. I am no Luddite - but folks I know in my industry have iPhones all right but believe they ought to use it only to make calls. So, in a driverless car, someone still has to give instructions or type in or whatever. And I dare say 70% of even millennials will be incapable of giving these instructions.

I’m disappointed that Mr Haberman didn’t even mention, much less explore, the role of government in bringing about the autonomous car. It’s as though he drank the same libertarian Kool-ade Google is selling. Great advances depend on government support. Moon landing, interstate highway system, the Internet: All are the product of direct government investment and direction. One comment here presciently cites V2V communication. Where is the tcp/ip research into vehicle communication standards? Obviously autonomous vehicles would benefit from communicating with nearby vehicles; in fact, human-driven cars would, too. If every stop sign and traffic light and off ramp included a radio beacon to alert/control cars, drivers would gradually be relieved of the duty for constant vigilance. When that technology is as ubiquitous as cell phones, the transition to autonomous vehicles will be safe and obvious. The mistake is trying to replace human drivers’ capacity with robotic cars. Play to the machines’ strength: incorporating information quickly and concurrently. Build an environment for them, that they can better serve us.

What concerns me about self driving vehicles, and the current assumption that a driver should be alert and ready to take the wheel, is not only reaction time, but skills. I grew up driving in winter conditions. I know how a vehicle handles in slippery conditions. I find it impossible to "slam on the breaks", or otherwise make sudden corrections. I think my vehicles, (Jeep 350k miles/Saab 230k miles) attest. Even with the advent of anti-lock breaks (which I have never needed to engage on the one car that has them), and the sophisticated anti-skid technology in newer vehicles, I still see people do stupid things. When our vehicles drive "for us", then those skills that have been honed will effectively disappear. That patch of black ice that momentarily sends the vehicle into a fishtail is going to effectively disable any automated system, and the driver will be helpless to correct. Granted, the new technology has probably prevented many serious accidents already, but I'd rather know that people still know how to drive.

I cannot wait for driverless cars. I have been rear-ended at least 8 times by distracted or impaired drivers. Yes there will be deaths and as soon as they happen you'll hear the refrain "See we told you so.." But you heard the same thing for many years about seat belts. Yeah, some people have been trapped in cars and died because of a seat belt. But ten of thousands of lives have been saved. According to the CDC, every day 9 people are killed and more than 1,000 are injured by distracted drivers. As quickly as we can, we should adopt the driver equivalent of a seat belt and let the computer take over.

It used to be step on the gas to go. And step on the brake to stop. Now it is an endless series of mystifying computer messages, instructions and beeps. I am completely baffled by the new cars. So are millions of other senior citizens. The difference between me and them is that I have quit driving, and they have not and are still out there with you.

There is no reason to believe we can prevent this technology from being hacked. None of the AI developers is explaining why autonomous cars would be different than all the other systems that have been penetrated. A couple incidents where hundreds are killed or injured due to a terrorist attack and the technology will be untouchable.

In 50 years people will wonder how it was ever possible that humans were allowed to drive cars.

Why don't the cities just work on synchronizing the traffic lights first?

No I wouldn't ride in a driverless car right now. I barely use cruise control and I hate being a passenger when someone else is driving. I think "driverless" has been really hyped by Wall St. investors, and Tesla is an example. Let's keep introducing technology into our cars and trucks. Back up cameras are now standard; let's make lane departure, blind spot detectors, adaptive cruise control (which is really cool) automatic braking and steering wheel monitoring standard equipment, too, as commonplace as seat belts. Let's keep evolving tech in cars, and spend less time and money (and hype) trying for the revolutionary.

The problem says a lot about how we constantly apply our preconceived judgements to incoming data. People are wildly inaccurate in their assessments of risk, instead fearing the things that evoke emotion rather than looking at the data. Even if self-driving vehicles become dramatically safer than human drivers, people will still vilify the technology when an accident does occur. Meanwhile, we give people a pass and routinely accept or ignore indefensibly dangerous driving, and even intentional harm, for example, to bicyclists by human drivers.

I’ve been driving a Tesla model 3 for the past 3 months and it does drive itself in limited conditions but always improving. Like with many other technologies it will take some time but it’s here already. It won’t be perfect but certainly better than the average human. That’s good enough for me.

Two years ago I attended the Intelligent Transportation Society's annual meeting -- a global gathering of scientists and engineers working on autonomous vehicles. During a public Q&A I asked one panel of Stanford and MIT PhDs, "Can any of you imagine a really large-scale deployment of autonomous vehicles in the absence of nearly-complete vehicle-to-vehicle connectivity?" IE, must V2V technology precede autonomy? The panel took that question as something that basically should have gone without even asking -- *Of*course* V2V adoption would so greatly simplify autonomous driving, that there was really no point in trying to develop autonomous systems in the absence of V2V. The reason I asked that question was that at that time, only Tesla appeared to be attempting to bring a self-driving car to market while ignoring V2V technology. All the other OEMs appeared to be operating on the assumption that widespread V2V would precede meaningful autonomy. Since then I get the feeling that other OEMs have assumed V2V won't happen, and are trying to follow Tesla's strategy of creating a truly autonomous car, ie, one operating on its own without input from other cars around it. That's a way fuzzier problem. If automakers hadn't lost their V2V momentum, we'd be safer in the medium term, and the long term goal of full autonomy would come about sooner.

While I can’t wait to get all the distracted drivers off the road. There is quite a bit more to this, the alignment of sensors on cars will add enormous technical challenges to the repair industry adding to the cost of repairs. And poorly repaired cars will be incredibly dangerous. I like to have my cars make as few decisions as possible.

I walk everywhere although occasionally take the bus. So no car at all. Recommended.

I drive on interstate highways every day in a large Metro area. It amazes me how many people, especially younger people, are constantly looking at their phones. I full support driver augmented vehicles, in fact, the sooner it's fazed in, and insurance discounts are offered if this technology is available, the better

Nobody living now will see a 2 ton minivan drive down a residential street with nobody in it. Software security alone forbids it. Nobody will insure it. Silicon running software is not up to the task ... they can't close the final 10% and everyone in the autonomous vehicle space knows it. Only glib exec selling IPOs to gullible investors still tout the possibility.

Longer than expected...really?

Yes, driverless cars will replace drivers who are paid to drive. Isn't that part of the point? The general idea of an economic system is to produce the greatest goods and services with the fewest resources. If a person can be replaced by a machine, that's a net benefit. That person is freed to do something else, creating an additional good or service. My butcher doesn't slice meat with a hand saw, gardeners don't use push-mowers and books aren't cut out of stone. I don't have to hand-pump water out of a well, either. Automation is a good thing.