Civil Maps, Here, Nvidia and other companies are working on a new kind of detailed cartography. Those new maps of the world will be for cars, not humans.
The weight of the automotive and tech industries is fully behind the move toward self-driving cars. Cars with “limited autonomy”—i.e., the ability to drive themselves under certain conditions (level 3) or within certain geofenced locations (level 4)—should be on our roads within the next five years.
But a completely autonomous vehicle—capable of driving anywhere, any time, with human input limited to telling it just a destination—remains a more distant goal. To make that happen, cars are going to need to know exactly where they are in the world with far greater precision than currently possible with technology like GPS. And that means new maps that are far more accurate than anything you could buy at the next gas station—not that a human would be able to read them anyway.
Fully aware of this need, car makers like BMW, Audi, Mercedes-Benz, and Ford have been voting with their wallets. They’re investing in companies like Here and Civil Maps that are building the platforms and gathering the data required. The end result will be a high-definition 3D map of our road networks—and everything within a few meters of them—that’s constantly updated by vehicles as they drive along.
Here started work building HD maps back in 2013, according to Sanjay Sood, the company’s VP for highly automated driving.
The first step is to create the initial map, which involves sensor-encrusted mapping vehicles that put Google’s Streetview cars to shame. Here uses a fleet of vehicles equipped with a roof-mounted sensor mast that packs 96 megapixels’ worth of cameras, a 32-beam Velodyne LIDAR scanner, and highly accurate Novatel GPS Inertial Measurement Units. These mapping vehicles drive around creating a 3D scan of the road and its surroundings that gets sent to Here’s cloud. From the cloud, that data is incorporated into a cm-accurate digital recreation of the real world.
“Starting last year, we’re essentially building the road network in order to have this map available for the first fleets of cars that are going to be leveraging this technology that are going to be showing up on the roads around 2020,” said Sood. “So we have to seed that ecosystem with a map.”
Maps will need to be constantly updated to reflect closures, road works, and other major obstacles.
Here has a couple of approaches to solving that problem. “We have hundreds of these mapping vehicles deployed around the globe,” Sood explains. “We have very in-depth relationships with many cities and regional authorities. So typically when there’s a large construction project happening, we know well before it even starts, and in many situations we can drive our vehicles there before the roads are open to the public.” That will allow Here to deploy updated maps (which typically arrive as 2km-by-2km tiles from Here’s cloud) to vehicles the day a certain road is opened or closed.
“How do you take all this heterogenous data and then make sense of it when you put it into a big data lake?” Sood told us. “That’s where a lot of the specialized skills that Here has come to the forefront.”
Not Just Graphics Cards
You may only know it for its GPUs, but Nvidia is becoming quite the player in this field, too, thanks to the company’s expertise with machine learning and deep neural networks. Those have plenty of applications in the self-driving world, and Nvidia is working with Here, as well as with TomTom, Baidu, and Zenrin, on mapping and cloud-to-car platforms. “HD maps are essential for self-driving cars,” said Jen-Hsun Huang, founder and chief executive officer of Nvidia. “Here’s adoption of our deep learning technology for their cloud-to-car mapping system will accelerate automakers’ ability to deploy self-driving vehicles.”
Nvidia’s technology—along with Intel’s—will also be required to slim down the bandwidth bills that will surely follow a crowdsourcing mapping fleet. “We’re developing a compute platform for the car, where the car itself does the change detection,” Sood said.
Problems Addressed
Every problem that readers throw at me is being addressed. I was not even aware of some of these companies until recently.
One reader commented something to the effect that cars will never be able to handle every conceivable road situations. That’s a ridiculously high standard, isn’t it?
I can conceive of lots of things. So what?
This is practical reality: Millions of long-haul truck jobs will vanish the moment level 4 is reached on highways. Every indication is level 4 for highways will be in place by 2020. Give it another year if you like, but competition ensures that outcome is soon assured.
Within a year of that event, millions of long-haul truck jobs will vanish. Taxi-related jobs will vanish as well, but likely at a slower pace.
Mike “Mish” Shedlock
this is all level 4
level 5 will not need maps
“where we’re going we don’t need roads”
I want the level that allows me to be dropped off at the entrance and let my car find parking for itself.
How about the next level after that? The car drops YOU off and goes to wherever YOU were going and does whatever TOU were going to do! At some point cars/machines/robots etc will decide “where we’re going, we don’t need humans!”
.
Mish did you happen to read how Uber vehicles are a complete
disaster – and how the truth is autonomous vehicles are still in their
infancy.
What a disaster how you and so many have been misled by the VC
money mongers.
In the US, traffic accidents are a leading cause of death for people under 25.
‘This is a tragedy that self-driving technology can help solve.’
But leaked documents have revealed that the cars are struggling to drive without human help.
Uber has been collecting data on overall ‘ride experience’ which is broken down into three categories.
These include: ‘miles per intervention’, how many miles the car can drive without a human taking over, ‘miles per critical intervention’, how far the car can drive without making an error that could lead to them hitting a person or causing serious property damage and ‘bad experience’, a measure of the overall smoothness of the ride.
Data revealed that in January, the cars could only drive for 0.9 miles (1.5km) before a human had to take over the wheel and in February this figure dropped to 0.7 miles (one km).
The number of ‘critical’ interventions made to avoid dangerous crashes has fluctuated wildly from week to week.
In January, the cars could go for 4.5 miles (seven km) without having a ‘bad experience’ but this number dropped to just two miles (three km) by February.
Uber has not commented on the leaked information.
Read more: http://www.dailymail.co.uk/sciencetech/article-4323396/Uber-automated-cars-drive-mile-without-help.html#ixzz4bb1hryAN
.
.
. Poor misled souls……
Uber’s automated cars can only drive for just less than a mile without a human having to take over the wheel, according to shocking leaked data.
The cars, which are being road tested in Pennsylvania, Arizona and California, could only drive for 0.7 miles (one km) without assistance in February.
And during one week, the vehicles could only drive for 50 miles (80km) before needing a ‘critical’ human intervention to prevent them from hitting a person or causing at least £4,042 ($5,000) worth of property damage.
Read more: http://www.dailymail.co.uk/sciencetech/article-4323396/Uber-automated-cars-drive-mile-without-help.html#ixzz4bb2tX4CQ
===========And Uber is spending billions upon billions
What a disaster. Poor misled souls……..
.
.
.
If any of them driverless cars can navigate Pittsburgh hills in the winter for a mile, it is doing better that the majority of Asian Females I see on the roads.
this is your “told you so”? it’s not convincing or shocking. imagine what the horse and buggy industry supporters said about “horseless” carriages. it may not be hard for you since you sound one in the same.
Makes sense. Another way to approach this problem would be to ask what it is that Google, the pioneer in automated driving knows that the rest don’t. Google has been working on this problem for almost 15 years now and even then its cars cannot drive without detailed maps of places around Mountain View, which is where they keep driving all the time. And they still don’t have an autonomous car that I can buy today. Worse, they scoffed at 100% self-driving car last year and explained why they think it is not feasible.
Uber’s story is even more incredulous. The prof from CMU who runs the automated driving at Uber first told Travis that what he was seeking is not possible (this is the prof’s narration). But Travis kept calling and after a few phone calls, the prof said it is possible! Just like that. This has Kramer (from Seinfeld) and his internship written all over it. If you want to go to Pluto, call me 30 times. The 30th time, I will say yes and we can all go to Pluto for a beer.
The key point is that even the QB prof at Uber did not think the kind of computer vision based driving was completely possible. Yet, suddenly everybody has solved the problem. IMO, driver-less cars being touted by Uber are in the same boat as VR — mostly hype (try out Oculus Rift and you will see what I am talking about).
I think the detailed maps will be needed to make the problem more feasible but I am not convinced it is cost viable, unless DOT jumps in and embeds sensors everywhere (more than what it has done now). I don’t see that happening in the current atmosphere when the stress is on putting people to work (and not robbing their livelihood).
Even assuming you solve the static issues like map granularity and up-to-date-ness, you still haven’t even come close to the Big Kahuna, which is autonomous agent interaction.
People driving on the road, is not some sort of giant borg hivemind attempting to solve a routing problem cooperatively. Instead, every driver is competing with every other for position. Constantly taking “calculated” risks to gain a few feet at the expense of the other guy. Risks to their own life, as well as those of others. Expecting robots to keep up with humans in that arena, is pretty darned futile.
23 year old MIT grads from sheltered suburbs with “you can accomplish anything you want, be anything you want, and everyone hold hands and sing Kumbaya” style “educations” can be excused for overlooking this. But you’d think the readership of a blog at least tangentially related to “economics”, would see it right away: Attempting to “solve” the inherently (in the current environment) competitive problem of driving, with centrally planned algorithms, are not really any different from solving economic problems in general that way. Central planning, IOW.
And an inevitable, scary corollary to that is: As the various high profile experiments starts failing by showing their limitations, their wealthy and well connected fragile-ego backers will solve that issue the only way they have ever known how: By havinh government limit regular people’s freedom of action, so that the illusion their committed to Utopia “works,” can be maintained. Berlin Wall style. As in: You can’t afford a RoboTesla? Stay off the freeway! So your erratic “lack of driving skills” stops posing statistically increased risk to the life of your overlords who can! Take the bus, pay your taxes and fines, work harder on that treadmill, and proudly wave a flag to show your appreciation for how advanced “your” country is! Rah Rah Rah Rah!
That’s not to say the tech being developed is not massively beneficial. Largely automating long haul trucking from freeway entrance to exit a few time zones away along a largely empty controlled access freeway, is infinitely easier than programming a car to respond optimally, or even acceptably, to being involuntarily engaged in a running gun battle through Chinatown while carrying a nuclear warhead in the trunk, that absolutely must not fall into the hands of the ones who did the engaging.
And, it’s not as if the transportation environment inherently needs to be, or even benefits from being, some free for all competitive field, rather than a more cooperative one better suited to automation. Controlled access freeways are a good example of the benefits of occasionally limiting absolutely unconstrained freedom.
But in order to not make the freedom restriction extreme, and very much for the benefit only of the well connected and wealthy, you can’t get to the benefits by attempting to overlay autonomous vehicles over the existing, built for humans, driving environment. As the cooperative and coordinated environment that is necessary for robodrivers to thrive, will inevitably be gamed to death by even the simplest of humans. Who will then be banned. To “pjotect the inveeestment vaijue” of the privileged.
Instead, build the autonomous corridors in parallel. Then people would switch over, once the benefits became obvious and cost effective for their specif situations. Not when some central planners decide their “skills” and “safety” is less than that of some robot, according to some “model” that some well connected “experts” babble about on TV.
As huge bonuses, this 1) allows the autonomous roadnet to be built on some form of rail as well. Allowing for higher speeds, lower maintenance etc. Rather than throwing all manners of resources at automation, while still being limited by rubber/rock friction coefficients and wear. And 2) lights a proper fire under the impetus to get rid of nimbyistic zoning nonsense that is preventing a parallel transportation network to be built alongside the existing one. Those societies and locales who pay the most heed to the nimbyots, will fall behind. As they are supposed to.
“Google, the pioneer in automated driving knows that the rest don’t”
Actually the defense contractors, particularly Lockheed Martin, Raython, Honeywell, etc., are dramatically more equipped to deliver self-driving cars than Google with their reputations and actual experience in systems engineering, sensors, and highly reliable embedded systems targeted currently at aerospace and defense applications.
Google is a joke, and its becoming increasingly obvious to the marketplace. Same with Uber.
Uber is a joke. I did a report on Uber. But Driverless will happen by 2020 in a limited way.
Technology is advancing rapidly. 2022 is a long time at the current rate of progress
Spoke with an engineer from the Google Waymo team the other day…he says they are decades away before this becomes reality, if ever. But they are all touting it in the Valley to get idiot investors bux.
Mish is too conservative. The moment a level 5 add-on is available for my car I will purchase it. I’d be willing to pay at least 5k to retrofit my current car with autonomous tech. I suspect I’m not alone.
I’ll be buying (2) level 5 retrofit kits the day they come on market
one for my 82 mom
the second for my 89 year old godmother
both are delightful and I love them dearly, but they drive like drunks with 0.9 BAC
the third retrofit kit will be for the family beater – little dude will be turning 16 soon
There will be no retrofit kits. Self-driving requires such fundamental and significant changes to the electro-mechanical systems in a car that any such vehicles will look dramatically different and most likely weigh significantly more than anything on the road today. And because of the mission-critical nature, require extremely expensive regulated maintenance.
Go talk to anyone who owns an aircraft about the sort of costs that are involved with maintenance, and scale that up by a significant factor due to the significantly increased complexity of a vehicle versus a light airplane. Most American vehicle operators can’t even be trusted to take their vehicle in for a regular oil change.
(raps gavel)
“The Luddite Skeptic Society meeting is now in session.”
Well I think they should have used green triangles instead of yellow ones in that pointcloud map. It isn’t just aesthetics but also a wider statement on the need to integrate technology into the tonal sensibilities of a suburban setting. It is also possible to tell that the maping engine has extrapolatory programming features that enable it to design hypothetical extensions. This is clearly visible in the background where the presumed road abruptly detaches from the photographic inlay and heads off into empty space, something that is not as disconcerting for a computer as a human being, except those that frequent Space Mountain that is. That would explain why these maps are deemed not for human use I would suppose.
That’s nice, but I haven’t seen anything in this about how the cars around you are communicating its states to the car you may be travelling in-kinda seems important to me if something happens for traffic flow control…..
Every time they solve a problem you will point out the remaining ones, until finally the mass proliferation of auto-cars is staring you in the face.
Drivers will embrace the tech. Commuters would, on average, rather sleep, work, read, eat, etc.
Parents would rather have their kids riding in a traceable pod.
Children of seniors would rather mom and dad not drive, and mom and dad would be more willing to visit grandkids.
Business travelers might make a 4-6 hour ride instead of a flight when factoring in all of the wasted time going through security, boarding, de-boarding.
Over 1 million lives will be saved yearly from auto accidents.
Millions of injuries and lawsuits will be avoided, as will wasteful expenditures on auto body repair.
Are you serious or is this sarcasm? You see, they have this newfangled technology called radios….
I can think of 3 ways to solve this “problem” just while I’m typing this response and I’m sure there are others. Aircraft like transponders, including the location of your vehicle in the data feed to the map, or something simple like incorporating special reflective surfaces on the corners of vehicles that are readable using specific light waves. (Something similar is used in military applications.)
I’m pretty sure he was serious. This is the same crowd who keeps harping on the idea that rogue hackers, terrorists, and bored teenagers will hijack cars routinely and crash them into each other. Yet none of them have the same fears of the massive amount of consumer goods available that could be weaponized, yet we’ve had one pressure cooker bomb attack since the inception of the pressure cooker.
And of course some alphabet agency will demand a back door so they can take over control of any or all “autonomous” vehicles, “in case of hijacking” / “to save the children” / “global warming” blah-blah.
Nearly every interstate has an adjacent railroad that doesn’t need any digitization, is non-stop, has the right of way, is 100x fuel efficient and requires 1 driver which might be zeroed.
Instead we need to spend millions on self-driving trucks to enable them do what the railroads do every day with lots of brittle technology and unknown risks.
getting rid of the RR engineer gets rid of 1 person
How many RR engineers are there?
How many truck drivers are there?
If RR are such a good deal, why do trucks still exist?
– Rail capacity constraints
– Shipments not going to a transit hub
– Transit times to less frequented transit hubs
It would cost billions with many court fights around right of ways and eminent domain that would take decades to sort out and implement and expanded rail system to take over for the trucks.
At the end of the day though it is irrelevant that RRs do a similar job, there are still lots of trucks for a reason and making them cheaper to operate, if anything, will pull more freight off the rail lines and make them less competitive.
Not to mention that railroads have schedules. You can’t just send a train out on a whim
That’s not to say that improving derailing tech to the point of increased competitiveness versus over the road trucks, is not ultimately a lower hanging fruit than automating trucks is on many stretches.
There is a huge (I’d venture to say ultimately more fundamental than “self driving cars”) move underfoot to build very granular “packet switched” physical goods transportation network spanning the world. Encompassing packets down to letter sized, and involving “in flight” transfers between airborne drones, trucks, and over time rail, planes and ships as well. Highly dynamic railroad derailing will result from this, regardless of whether passenger cars make use of it. Allowing an order of magnitude or more improvements in rail utilization, and reduction if loading/unloading times.
Once this current version of Google-recruitment-myth motivated AI hype inevitably turns into yet another in a never ending line of AI winters, you’ll even start hearing about it outside of Amazon and the like.
As a person who has spent a little time in the logistics business using trucks, trucking
was never less expensive than the RR. What the customer was willing to pay for was
specific times of PU and delivery, possible manpower or storage on the van waiting near the dock. Local trucking and drivers will not be replaced by this proposed technology and
requires the same level of scheduling as it always has.
If the timing was not an issue or dock storage was lengthy, we put trailers on the rail.
Railroads called such trailer loads-“hoggers”. USPS stopped using the RRs and used
trucks, they have since put the long haul trailers back on the rails. Please review the initial
picture as it looks to be a 2D version of railroad tracks placed on the freeway? A far
cheaper money saving solution would be dedicated truck lanes and dedicated/limited
trucking offramps on all major through ways.
Apparently building autonomous cars is an easier problem to solve than reforming regulation and modernizing the rails.
I expect dense cities to work, but even better/worse to put new “no turn on red” or “no right turn” changes which will take a while to be mapped and raise millions of revenue. Hey, maybe enough to save Chicago.
There is near zero security – like the Jeep that can have it’s engine shut off over the cell connection. It is one thing if there might be a human witness, another entirely if someone can create autonomous vehicle ransomware.
This article was hilarious.
I don’t want to get into Mish’s obsession with self-driving cars again. I want to draw attention to the absolutely absurd claim from “Here” corporation that they will know when construction projects start and end — often before anyone else.
Bullsh!t.
Anywhere in the northeast USA, this claim is just insane. The DMV doesn’t know what is happening at least 50% of the time, the other 50% of the time the state DoT doesn’t know — and there are plenty of cases where neither is in the loop. And as any person leaving in the real world knows, it is not unusual to call the same 1-800 “public service” number for a state three times in a row and get three completely different answers.
Anyone with real world computer experience (I am talking a business computing course, not the hard core programmers) has heard the saying: garbage in, garbage out.
If the data going in is bad (and on construction projects, it most assuredly is crap) then the data coming out is also bad.
Do I believe the lazy dolts at the DPW / DMV / state DoT are suddenly going to start communicating with each other? Or will they continue focusing on retirement with full pensions after 20 years of doing nothing?
If you believe state employees are going to communicate with each other harmoniously, never mind communicate with any outside party — than I want to sell you ocean front real estate in Montana.
Admittedly, I don’t have as much experience in states outside the northeast — but a quick glimpse at Mish’s neck of the woods (near Chicago supposedly?) tells me that chaotic road construction problems are not just in the northeast.
How many years did Boston’s “big dig” go over schedule? And quite a few sections needed repairs within weeks of being “finished” — so one could argue the bureaucrat that claimed the project was finished didn’t communicate with the other bureaucrat that said “repairs already needed”.
Many parts of I-95, at least the sections between Washington DC and Boston — have been under perpetual construction since they were “finished” in the 1960s.
Here Corporation has no idea when these projects will finish… they are bluffing, and not even a plausible bluff.
Yeah no kidding, and where’s the trillion+ in road upgrades going to come from to make self-driving ‘work’? The government is broke. And with such a limited number of self-driving vehicle users, they’re not going to be able to pass the costs entirely onto the users.
DMV’s and DoT’s can barely, if even manage the civil infrastructure they already own. I can’t possibly fathom them getting into building and managing the sort of infrastructure needed for self-driving vehicles.
We’ll all be dead by the time such fantasy is operational.
I never liked science fiction.
When I was reading Mish’s post, I was thinking of the movie Jurasic Park, where the park developer kept babbling “we have multiple redundant systems that are infallable, the dinosaurs can never escape and never reproduce…. we spared no expense!”
Meanwhile Newman (I can’t remember the actor who played Newman on Seinfeld and the computer guy in Jurasic Park) was busy sabotaging the computer systems while the dinosaurs were having babies in spite of the infallible controls.
Its funny when you see this stuff in a movie. But when a real world corporation makes ridiculous claims that they know what the transportation bureaucrats are doing before the bureaucrats… the bureaucrats don’t know themselves, and can’t communicate among themselves.
These won’t be self-driving cars, they will be government programmed cars. I certainly will never own one, not even if they give it to me….Since when has the government ever done anything right, or cared about casualties to the public?
I just wonder what will happen when someone is in traffic between two loaded cement trucks and a deer runs out in front of the first one. Think about it.
What would a driver do? A computer would keep a safe distance from the vehicle ahead. It would also apply brakes quicker. Humans are dangerous behind a wheel. Irrational and easily distracted. Computers are safer. If you don’t believe me, ask an auto insurance company. They have the statistics to back up their answer.
What happens now, with human drivers?
all three vehicles crash and burn because the humans drivers are driving too fast for the conditions and are texting and are day dreaming
bambi gets squashed
authorities consider the carnage just another ‘accident’
Right now, one of two scenarios happen:
(1) smart drivers do not put themselves between two dump trucks, even if there are no deer etc. People who were educated in high school physics years ago (not the gender inequality crap that passes for education now) …. high school physics courses years ago taught even the half wits that a car has a vastly different stopping distance from two cement trucks. We also learned that, sitting in a car, you can’t see past the cement truck in front of you.
(2) liberals like Mr Greenberg will get sandwiched between the two cement trucks, and then he (or his heirs) will add further insult to taxpayers by filing a lawsuit against the cement trucks and against the DoT (aka taxpayers). The cement trucks are driven by corrupt union guys, or by illegal immigrants that Greenberg insisted had inalienable rights to screw up the country they were squatting in. Either way, the cement trucks won’t pay Mr Greenberg’s hospital bills (illegal aliens will flee the scene, union dopes own the judge/courts). So the taxpayer will get hit with a massive lawsuit because Mr Greenberg doesn’t grasp physics and would rather offend taxpayers instead of illegal immigrants.
And that is why Trump got elected instead of yet another crooked political insider…
I’m not a liberal you ad hominem-slinging douchebag.
The liberal doth protest too much…
And I don’t hide behind synonyms fuckface.
Not sure what they are teaching in gender inequality studies these days, but a synonym is a word having the same or similar meaning. Medex Man is a ***PSEUDONYM***, not a synonym.
Blue collar guys learn about the physics of cement trucks versus cars. White collar guys learn extended vocabularies including synonyms and pseudonyms.
Liberals learn about safe spaces and money that falls from the sky to pay for all sorts of corrupt welfare nonsense (the poor are still poor after all the welfare is taken up by the bureaucrats and non-profits).
Just because it’s possible doesn’t mean it’s a good idea.
Speaking purely logically, just because you think something’s a bad idea does not mean it’s a bad idea. But your opinion is noted.
If history tells us anything it is that once control is given, it is extremely difficult to get it back. Our own government is a good example. There is a price for our conveniences that we are just now starting to realize.
What happens now, with human drivers?
Mish…one major concern and danger of autonomous autos I have not yet heard discussed is the inability to control the route of travel…how do I prevent my auto from steering itself into and through the equivilent of South Central L.A….There are unsafe neighborhoods in every city in America…main access is blocked? Hope you are right with your diety because we are taking a tour of the shooting gallery…Safety be damned, MPG and carbon footprint are more important than the loss of six quarts of your blood……and here’ s the kicker…..wait until the ACLU discovers that auto companies are red lining these neighborhoods ….you’ll be a racist/bigot full of predudice and no compassion for the less fortunate in life…The cops will want a piece of the action too, with an undercover on every corner soliciting new felonius business from unsuspecting riders…..As for me….I can still ride my motorcycle…
Ever tried navigating using Google Maps? It offers you several routes you can choose and the time associated with each route. You just tap on the route you want and voila! It navigates for you, the only difference here will be that the car will actually drive as well
Interesting point. I can see the car manufacturers selling your eyeballs for extra profits. Say you have multiple paths to a destination. The car takes you along the route that has the most profitable advertiser’s billboards. Maybe with a message that comes in through your radio along the way. The car might even slow down so you can get the full effect.
Of course, nobody will want this intrusion. But for some reason, there won’t be any car companies producing vehicles without these bonus features. Unless you pay an extra monthly fee of course.
So if McDonalds started serving Big Macs with veggie burger patties and called it beef, would you continue buying them? No, you’d go to Wendys.
If Google or whoever is routing your car started pulling such shenanigans, you would find another route service provider.
And If I could go to a website without advertising, I would.
I wrote a short story that used that exact idea as the main plot point. The idea was that the nav company would route you past paying stores and restaurants, leading to some roads that were in perpetual gridlock while others without businesses were empty. The main characters made a game out of manual navigation.
“One reader commented something to the effect that cars will never be able to handle every conceivable road situations”
As if any human could either…. Most people can’t even parallel park properly
Accidents due to speeding will be a thing of the past. Tick.
Standards will be set for car-car communication so you system will see a few cars ahead for emergency breaking etc.
Intel bought mobileye, recently spoke to someone who had seen the data off a system. Say no more. Very impressive capabilities already.
Our roads and bridges are in need of repair and we are spending money on driverless cars and trucks. Good luck navigating all the cones for closed lanes in cities like buffalo and Cleveland. Northern cities roads are terrible and in constant maintenance.
It’s hard to generate a lot of hype around repairing infrastructure.
Drones?
“We” are not spending anything. Entrepreneurs are trying to find solutions to a massive problem: 1M annual auto deaths, millions of crashes causing injury and wasteful repairs of cars (breaking things to repair them isn’t a positive economic activity), freeing up urban space, shortening commute times, and so on.
Perhaps the governments who are charged with maintaining these roads will start operating more like the entrepreneurs who are working on autonomous cars, and fix the roads with the fat budgets we already send them.
Out of curiosity are ‘bad maps’ the leading cause of failure in current self driving tech?
We don’t need to know ‘exactly’ where we are with precision at any given moment in order drive – driving jokes aside.
Why should these cars?
These maps are useful as a key element when much other data is obscured. For example driving after a big snowfall. The road and paint lines are obscured. So are many reference points like where the road ends to each side. But trees, signage, etc are still visible and by comparing to a map your vehicle still can navigate. Designs that follow paint lines are inferior to this. These 3D maps are so detailed and so unique place to place you can identify exactly where you are anywhere in the world once the mapping is complete. Way more reliable than GPS.
There is also technology to map the road substrate (beneath the surface). That technology can read through several inches of snow.
Lost of things in progress.
I love technology, it seeks to take the risk out of life. The problem is that risk requires exception handling. When things go wrong what is the best way, the possible way, the least risky way to handle the exception. Processes and procedures work until there is an exception, then they stop working and intervention is needed. The biggest problem with exceptions and their handling is that the major approach is to develop even more processes and procedures to handle the exception and hope that one gets it right. And just like Murphy’s Las, if anything can go wrong, it will. There will always be a new exception which will stop the whole process.
I wonder what will happen when malicious code is injected into the network of autonomous cars and they become Lemurs? Think about it
More reason for more wars!
“Within a year of that event, millions of long-haul truck jobs will vanish. Taxi-related jobs will vanish as well, but likely at a slower pace.”
https://www.bls.gov/oes/current/oes533032.htm
In 2015 there was about 1.678 million long haul truck drivers. Let’s assume within a year every single truck in the US will be instantly upgraded to be self-driving (unlikely, did tape players disappear from cars the moment CD players became available? Did VCR’s all vanish the moment DVR’s became available?). You only get ‘millions’ of jobs disappearing by rounding up. Even there you’re not really right.
350K drivers are owner operators. So if you own a truck and drive it your job is really running a very small trucking business. You could hire a driver but you probably just ‘hire’ yourself as the driver for the business. If you install a ‘self-driving’ system on your truck your job is still running a very small trucking business of one truck. Take that away from the 1.67M drivers figure and you can’t even get to millions by rounding.
Let’s also consider what it would cost. How much would a company charge for a ‘self-driving’ upgrade to a truck?
I’m guessing $100K+. I get that by guestimating that hiring a driver costs about $50K a year plus a self-driving rig could drive twice as much since you wouldn’t need sleep breaks.
A computer company could therefore advertise that the self driving rig would ‘pay for itself in a year’. when I search google for the phrase “Pays for itself in” I get responses that range from a low of 6 months to 3 years or so on average.
For owner-operators many would have a hard time immediately financing $100K+ (esp. since part of that ‘pays for itself’ would entail them having to find an alternative job). Even for trucking companies with a fleet, it would require serious financing to upgrade multiple rigs at that cost.
I’d expect the jobs to stick around for at least 5 years with most gone by 10.
I’m guessing more like $1-2M per unit for the “self-driving” hardware/software and necessary mechanical/electrical redundancy upgrades. Plus the maintenance requirements on ‘self-driving’ vehicles will be like that of aircraft — where you need highly credentialed mechanics performing the maintenance, supervised by engineering professionals. Along with tightly regulated supply chains for spare parts and various rotables.
Also, there will be significant support costs on an on-going basis from the software vendors themselves. And always a significant risk of the regulators simply deciding that certain self-driving platforms are technologically obsolete and must be decertified. A huge business risk to anyone who owns fleets of these vehicles.
Because self-driving vehicles will inherently weigh more because of their additional required electro/mechanical redundancy, useful payload will be reduced and fuel economy will fall as well. Additional strikes against the economics of self-driving vehicles, including trucks.
Fleet owners are really going to drop millions of dollars per truck for all this technology, instead of just employing a driver? Really? The problems are not particularly technological problems (although there are lots of technological problems which haven’t been addressed), but are heavily economic problems. Like who is going to pay for all the new infrastructure, trillions of dollars, worth, to actually equip roads for ‘self-driving’.
Also, if even, say, 10% of the current OTR drivers are laid off due to technology, will the price for the other 90% collapse? What will that do to the economics of self-driving when you can get a driver for $50k instead of $100k/year? As it stands the economics are bad enough (spending millions of dollars on a truck to replace a ‘driver’ that may still be needed anyways as a loadmaster/maintainer/etc.), but will be made even worse due to the deflationary impact of such.
My $0.02…
I suspect like most things, there will be phase-ins.
In my grandpa’s day horse drawn vehicles worked side by side with motorized vehicles … and unless there is a government mandate of the nature of “thou shall no longer use human drivers” folk will use whatever technology gives them the most bang for the buck that they can afford given their current commitments, debt load, income, etc.
Best wishes…
I agree that self-driving cars will need better maps, although how much better remains to be seen. 10 cm might not bee enough. I would go to the centimeter, but in that case, should we have a better positioning system ? GPS has some limitations and what about tunnels ? 3D positioning with no radio signal able to pass through 10k feet of rock might lead to the development of beacons, maybe next-gen Bluetooth beacons.