Slashdot Mirror
25 Percent of US Driving Could Be Done By Self-Driving Cars By 2030, Study Finds (techcrunch.com)
An anonymous reader quotes a report from TechCrunch: Self-driving still seems to be a ways off from active public use on regular roads, but once it arrives, it could ramp very quickly, according to a new study by the Boston Consulting Group. The study found that by 2030, up to a quarter of driving miles in the U.S. could be handled by self-driving electric vehicles operating in shared service fleets in cities, due mostly to considerable cost savings for urban drivers. The big change BCG sees is a result of the rise in interest in autonomous technologies, paired with the increased electrification of vehicles. There's also more pressure on cities to come up with alternate transportation solutions that address increasing congestion. All of that added together could drive reduction in costs by up to 60 percent for drivers who opt into using shared self-driving services vs. owning and operating their own cars.
Larger cities need to invest in mass transit and maintain it. Look at NYC, Tokyo and London as working mass transit systems. Smaller cities need working bus systems that aren't starved for money in order to be useful.
And I'll get ahead of it here: mass transit needs to be properly funded in order to work properly. Mass transit does not appear to pay for itself on the surface, it pays for itself because of increase in population density that occurs as a result.
Architectural plans are like computer source code with a couple of differences: You only compile once.
I think that number is very, very low.
My prediction is that within 10 years, half of new cars will have some level of self-driving ability. High-end cars will be almost all autonomous capable.
I also predict that 1 or more of the classic "big 3" auto makers will go under or be purchased.
I'll be toward the front of the line when cars have reliable self-driving capabilities. But I'm not sharing my car - that's personal space.
I like having my stuff (umbrella, bag full of fitness clothes and shoes, kids toys, pens, sunscreen/lotion, med kit, sunglasses, etc.) right where I want it at all times. I also like being able to clean my car to my standards and know that someone else hasn't been doing who knows what in my seat ten minutes ago.
So when I'm in a city and I need a taxi, I'll rent your shared car...you just can't have mine.
For at least 5% of cars on the road they already seem to be driving themselves.
Those self driving car engineers need to step up their game because the weaving side to side and driving well below the speed limit in the passing lane on the freeway is a dead give away that their algorithms can be improved.
I am Slashdot. Are you Slashdot as well?
We already have pooled driving and shared cars. It's called a taxi.
The only thing a self-driving vehicle does is take out the cost of the human driver. That's it.
People also carpool. That's been around forever.
Self-driving vehicles will change a lot of things: delivery trucks will go cross-country without sleep breaks, off-site parking will be more practical, highway deaths will drop like crazy - but nothing about city traffic will fundamentally change.
Ha! wasn't there just a Slashdot story about map software spoofing?
This will end well...
I have various health issues.
I cannot legally drive.
A self driving vehicle would be enormously freeing.
"I don't know anyone who has a self driving car or who has any intention of getting a self driving car. "
Of course. They are not on the market yet.
But now that the many beta models cruising with regular traffic are already better drivers than humans, it has become inevitable that they WILL come to market.
Let's say that trucking follows the same metric, 25% self driving inside of 15 years or so.
The trucking industry employs 3.5 million people. Source.
That means that we are potentially looking at 875,000 freshly unemployed truckers over the next 15 years.
Is anyone planning for this?
Weaselmancer
rediculous.
Mass transit does not appear to pay for itself on the surface, it pays for itself because of increase in population density that occurs as a result.
Sort of. Mass transit pays off biggest in cases where it allows you to use three dimensions instead of two. Subways allow you to use trains underground or overhead instead of on surface streets. Aircraft allow you to fly above the surface streets. When you get a dense city like NYC or Tokyo, you have people living in three dimensional buildings (high rises) but transiting in a two dimensional road network. This ensures congestion if you don't have a robust subway and tunnel network.
Busses obviously don't work in three dimensions but they can help optimize the 2D surface street usage. But above a certain population density they are limited in how much they can help. Street cars are basically trains on surface streets so they lack the routing flexibility of buses but don't take advantage of three dimensions like subways do so they really are a terrible solution for most circumstances as public transit.
People living there receive more services, and businesses located there get more customers
That depends on the services you are seeking and what sort of customers you are looking for. Good luck finding a tractor repair store or any customers for one in downtown Manhattan.
And nobody's forced to use a mode of transportation they dislike - you're allowed to walk, you can take a bus or train, or you can drive.
You can only drive in a city like NYC if you are rather wealthy. Costs too much and is far too impractical for most people. You essentially are forced to take public transit and not everyone likes that.
It's a win-win for everyone.
It's a win-win for people who want/need to be in a dense city. It's a huge loss for those who dislike living in such a place. Dense populations have their good and bad features. It's not a clear "win-win".
Forget the 3,000 lb. "missiles" that they want to control. If you want something real for humans, build a bullet train! Instead prove A.I. is trustworthy with smaller things first: Make automatic cooking and cookware cleaning. Make something you can drop on a toilet to convert it to an automatic butt cleaner, a shoe tyer for old folks, personal car washer and maybe a waxer too, an actual lawn mower that mows straight rows and is safe to children and pets and mows around obstacles, an automatic weed puller, a mail getter, a dog walker, a real home floor mopper that changes it's own water and puts itself away, a duster/crumb picker upper (Tell me that would not be popular!!), a real a.c. vacuum cleaner, a real animal feeder, a rug cleaner that cleans up animal droppings.. These a but a few things that are actually useful to the average person. A.I can actually make things easier for daily life for a change!
No, no it isn't. The sensory tech to make cars self-driving is already there. Google has millions of miles of successful testing under their belt. I myself have been riding in a Tesla going 120 km/h while the driver was doing nothing. Now Tesla's solution is not fully automatic and requires you to still pay attention to the road, but this if anything should point out to you that we're much, much closer to self-driving cars than most people think. There are issues to solve still before the tech is ready to hit the market widely, but these obstacles (handing of weather conditions etc) are things which can be solved with existing technology and innovations.
I will get one as soon as the price is right. I'd much rather climb onto the car, tell it to get me to wherever, and sleep/eat/do work/whatever else with my time rather than drive,
All existing data from ongoing self-driving car pilots, including Google's, point to self-driving cars being vastly safer than human piloted vehicles. The cars have technology allowing them to have a far superior understanding of their surroundings than any driver ever can. Google is using radar. The car can see and know the location of vehicles, pedestrians and obstacles far outside the reach of a humans vision. The car is able to do real-time physics calculation and actually figure out the smartest way to for example evade stuff. Humans take about a second to a second and a half to even react to a sudden emergency. In that second the computer has not only detected the obstacle/danger but processed through the different alternatives it has (ie. it has figured out whether or not to do perform an emergency braking maneuver or whether it's better to evade, and in which direction) to respond to that and has started to take action far before human reaction times even kick in. Literally in the time it takes for the human brain to go "OH FUCK A MOOSE!", the self-driving car has already mapped a solution and began implementation taking into factor numerous things (brake-force per tire, road surface conditions, etc) that the human never could. And if a crash is unavoidable, the car can calculate a way/place to crash the in a way that does minimal damage to the passengers and start dialing emergency services immediately, even rely to them crash data so that the dispatched emergency staff will know what kind of injuries to prepare to treat.
Computers are simply better suited to handle the chaotic nature of traffic. It can track and predict the movement of all of those vehicles and pedestrians simultaneously. This video showcases that well at around the 12 minute mark: the cars are all stopped at a crossing, and there's a cyclist that's about to blow through a red-light. The human-operated vehicles all miss this and start to move forwards on a green light as the cyclist hits the crosswalk, nearly hitting him. The self-driving car has known (based on speed and trajectory of the cyclist from LADAR) that the cyclist is going to blow through the red light before he even does so and doesn't start to move until the cyclist has crossed the road safely.
Couple that with the fact that the cars can communicate with each other. A car that spots something like a tree or an accident on road can relay this info to other vehicles which can take this into consideration immediately and re-route far ahead of time.
Yes, but those can (and will) be sorted.
"It is the business of the future to be dangerous" -Alfred North Whitehead
The vast majority of Boston driving could be automated today.
Current AIs could easily handle merging in without checking whether a car is there or not, getting in minor low-speed accidents regularly, and constantly sounding the car horn.
#DeleteChrome
My prediction is that within 10 years, half of new cars will have some level of self-driving ability. High-end cars will be almost all autonomous capable.
That's not exactly putting yourself out on a limb. That would be true if half of new cars got lane departure monitoring or adaptive cruise control. Like most new technology it's going to move both faster and slower than most people think. You'll see self-driving tech appearing in some vehicles but it's going to take quite a while to become ubiquitous. Some niches will probably move faster than others. Liability issues will hold things back. And development cycles are rather long in the auto industry and that isn't going to change.
Here's why it's going to take a while. Development cycles on new cars take 3-5 years and once in production cars don't change drastically for 4-8 years. Given that most cars under development right now do not have any meaningful self driving tech in them, you aren't going to see it start to seriously take off for another decade. I know this because I work in the industry and I know a fair bit about the cars we'll see in the next 3-5 years. Development of a car is a much more drawn out affair than most people realize. So you won't even see the tech start to arrive in a substantial way for another 5-10 years and it's going to take some time to get into the market place. 15-20 years is probably a more realistic time line to see mass adoption of self driving tech. There will be some niches that move faster but it's going to take a while.
This all assumes that the liability issues are worked out in a timely manner. It's not entirely clear how fast they will get sorted. It's also not clear how fast people will become comfortable with the new tech even if it works well and can be made available at an affordable cost.
I also predict that 1 or more of the classic "big 3" auto makers will go under or be purchased.
Always possible but you aren't basing that on evidence unless you are talking about FCA which has been bought and sold three times in the last 20 years. It's pretty unlikely Ford or GM will get bought because there simply aren't a lot of potential buyers given their size. It would require either some sort of catastrophe or some company like Apple with ludicrous amounts of cash decides to buy them. (Apple could buy both Ford and GM in cash if they wanted to but it would be a terrible idea for them to do it) Mergers on the other hand happen all the time. I could maybe see Ford or GM merging with another car company but otherwise it would take a company like Berkshire Hathaway to buy them and so far they haven't been interested.
It would also improve the quality of life for a lot of elderly patients with various forms of mild dementia.
My elderly mom is fortunate that she still has her wits about her and can still handle driving physically - but I've seen her get flustered, and can see the eventual writing on the wall. I've seen other people wrestle with when to take the car keys away from their parents... it's not something I look forward to. Self-driving cars with some sort of limitations regarding where it's allowed to go - perhaps a pre-programmed list of destinations - would be a Godsend.
#DeleteChrome
The year 2030 is when all the baby boomers are supposed to retire, retirees will out number workers, and two-thirds of the federal budget will go to Social Security/Medicare. Taxes will have to go way up to pay for everything else. Someone will need to drive all those seniors all over the place.
People die by the thousands on the road every year. Sure there are challenges but it's going to happen. I think in the next 5 to 10 years you'll see it begin. The testing has shown it's safer than letting people drive. Already there are automatic anti-collision features in cars.
Self driving cars will be a neat exhibit in Disney Land, like the "Home of the Future".
Cheap storage VM.
Most people would drive forwards without turns for more than 25% of their journeys. Self drive vehicles should be able handle this reliably by 2030.
are already better drivers than humans
Problem is this is based on data that is hardly that rigorous scientifically, from parties that have a vested interest in a specific result.
On the data from autonomous driving:
-Sometimes includes 'simulations'
-Necessarily done only on cars less than a couple years old
-Additionally, they have particularly rigorous maintenance
-Has the benefit of all modern safety systems apart from the autonomous driving system itself.
-Any time the autonomous system determines it might not be able to operate safely, it surrenders to the human
-Only done in select areas.
-Sometimes has a 'no true scotsman' tilt to it (e.g. 'autopilot' isn't an autonomous system when there's an incident, but sometimes it counts when it looks good)
This data is then generally compared with incident rates from cars that:
-On overage, are an order of magnitude older
-With frequently spotty maintenance
-Many don't have anti lock breaks, stability and traction control, adaptive cruise control, collision warning, lane departure warning/avoidance
-Operates under circumstances the autonomous system won't even try
-Done everywhere
-Contain willfully reckless drivers, who will be a source of incidents unless they are *forbidden* from driving, even if their cars *can* be autonomous.
My suspicion is the safest is the 'semi-autonomous', systems double checking the human and having a much wider view, but still with a human forced to pay attention will yield the best results. Note that systems that don't punish a user for neglecting to be attentive will give it a bad name (e.g. the Tesla incidents).
XML is like violence. If it doesn't solve the problem, use more.
All existing data from ongoing self-driving car pilots,
All publicized existing data from pilots run by groups that would profit from a favorable result, and generally interpreted by those groups into a press release that is regurgitated by an ever less critical media. Skepticism is not such a bad thing to have when faced with neither research nor journalism with an objective, critical eye.
XML is like violence. If it doesn't solve the problem, use more.
You're entirely correct. However, in general the reasons for the cars being safer (reaction times, better situational awareness) are such that they can - and must - be objectively demonstrated before mass market release. That is, we're not going to take the companies' word for it, obviously the cars need to be tested by outside/3rd parties before being put on sale.
My general point is that there's a whole host of reasons why computerized monitoring of traffic is safer than relying on a single human. I don't really see a logical reason for why a self-driving vehicles wouldn't be less-error prone simply due to the fact that human drivers make a lot of errors as is. That is, the bar for being a better driver than even an above-average driver is not nearly as high as people think.
"It is the business of the future to be dangerous" -Alfred North Whitehead
no stop signs & stoplights = underpass / overpasses for pedestrians.
" self-driving electric vehicles operating in shared service fleets in cities", that's a description for electric, autonomous taxi's.
Taxi's are here now and under used. Electric vehicles are here now, but the range sucks and they're not practical for 1/2 the country. Try using a battery vehicle in Minneapolis in January - nope.
Once self driving cars are widely available and "safe", which could be (c) 2030, traffic will double. All these prognosticators ignore American / human nature. We like to drive alone. If you have a vehicle that drives you to work, why pay the high cost of city parking when you can send it home, have it go pick up the kids after school, have it get groceries (Walmart is already planning for this eventuality), release it into a revenue earning driving system (send it to work for you in Uber while you're at work), have it go run any number of errands for you- but it still has to come back to get you after work. Now instead of 2 runs (1 to work, 1 home) there are 4 runs + errands all of which will effectively double+ traffic in cities. Sounds like more of a problem than a solution.
At 25%, self driving will come nowhere near to meeting all the lofty safety estimates being thrown about. Nor is it enough to offset expensive insurance costs so that people in self driving cars get affordable insurance. This is very sad news for self driving indeed. How long until we are at 95% because that is probably closer to the critical mass we need for self driving to actually work for common people.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
I see so many features that have been available in cars for years and not available in low end vehicles, I really wonder what could possibly trigger automakers to decide to put something as advanced in a low end vehicle. They had to be forced by law to put airbags and seat belts in, and those cost what, $10 per vehicle?
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
Or, they might not.
However, now that buffaloes have wings, you might want to keep you head down!
Sent from my ASR33 using ASCII
no stop signs & stoplights = underpass / overpasses for pedestrians.
Or walk signals for pedestrians, which tell the vehicles to stop. But, yeah, when intersections have free-flowing traffic except for pedestrians there will be a lot of incentive to build pedestrian overpasses or underpasses.
Note to ACs: I usually delete AC replies without reading them. If you want to talk to me, log in.
While we're running comparisons, as long as self driving cars don't cause more accidents than other heavy machinery on the market currently do, automated or not, then I'll be happy. Let's be clear here that companies are selling a machine for the purpose of transporting people, it should be held to current industry safety standards for heavy machines.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
Maybe it could look like the 80's or 90's. It would be a nice improvement to the Tomorrowland that looks like the 70's.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
Self driving cars are only safe if a majority of people on the road are driving them. That will be well beyond the time span in this article, which only gets us to 25%.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
I'd rather walk everywhere the rest of my life than ride in some so-called 'autonomous' death-trap.
Why are you using future tense? That lawsuit happened before you were born; people have been getting killed by cars for a century. It shouldn't surprise you, because one of the problems cars had from the beginning, is that they were operated by apes, and the apes had atriocious safety records, killing tens of thousands of people every year and causing immense property damage. Been there, done that.
It's not about whether or not you trust something, as it's about how much you trust it. This tech is only competing with the apes. It's merely vying for a 1st place win.
Yes, there are going to be some horror stories (some of them amusing, to those of us who appreciate black humor). People are going to die because of math or logic errors. Innocent children carrying cute puppies in their laps, sitting next to nuns, are going to slowly burn to death, screaming, due to known bugs that didn't get patched! Someone you love might be maimed for life because someone decided it would be fun to trick a different car into driving into oncoming traffic. That cool guy you know, the one who always seems to do the right thing, will be killed because some middle manager told his engineers, "That's not a common case, we'll handle it next revision." There will be traffic jams that the robots don't know how to get out of. Illusions might be able to trick sensors into parking stupidly.
Every bad thing you're imagining: it really might happen!
But take a look around at 2017. You're living with all that, right now. All those horror stories are the status quo. The goal is to have less of it.
In what universe is 13 *years* just to hit 25% "very quickly?" We need to put a plan in place that *prohibits* humans driving cars at all. 2030 seems like a good goal for that. Only then will self-driving cars realise their full potential, in terms of safety, efficiency, environmental savings. Individuals need to be prohibited from purchasing vehicles, unless they can justify the expense for a work-related reason or living out in the middle of nowhere.
There are things that inarguably are going to work better in software and sensors (the systems can know at the same time 360 degrees and within inches of each corner, if appropriate equipped) and can manipulate hardware faster when correct decisions are made.
On the flipside, we have whether the systems fail to recognize something real or 'hallucinate' something fake and react poorly. If some completely unexpected situation comes up not in the training (while driving training may be more exhaustive for the system than for a human, the human driving training takes for granted a ton of not-strictly-driving but still relevant decisions).
Also we frame the debate between completely manual and completely autonomous generally being the only two points to compare. In reality, we already see some of the most valuable safety and convenience of autonomous driving working itself into market ready cars (adaptive cruise control, lane guidance, collision warning/auto braking) that takes a lot of the tedium out and uses computing systems for things they are naturally good at (paying unending attention to detail that would bore a human brain to death) and the human is providing the ability to intelligently take in some of the bigger picture and make decisions. It's not such a bad place, and that should be the comparison point for autonomous driving, rather than people driving around 10 year old cars.
XML is like violence. If it doesn't solve the problem, use more.
A self driving vehicle would've helped both of them out immensely.
hey, so a few million kids pets and cyclists die, what's the diff?
-- Tigger warning: This post may contain tiggers! --
... is as American as apple pie.
What's a car, grandpa?
-- Tigger warning: This post may contain tiggers! --
These technology and dollar-and-cents kind of analyses always leave out the most important thing: human psychology. The acceptance or rejection of control loss will be the deciding factor. The rest doesn't matter if people refuse to sit helplessly in a vehicle that does what it does in some unknowable way.
E Proelio Veritas.
Considering how rapidly this tech is advancing I believe it'll be introduced withing the next 10 years. It might be many years after that but eventually almost all cars will self drive. Who knows, one day maybe only law enforcement vehicles will be allowed to drive manually, if those.