Robo-Car Drives Across Country 99 Percent Autonomously
Congratulations to Delphi on this feat

In 1919, the U.S. Army drove a motor convoy from Washington, D.C., to Oakland, California. The roads were so bad that driving at about 6 miles per hour, it took around a month for the convoy to cross the country. This historical crossing led to the development of our National System of Interstate and Defense Highways.
The car parts manufacturer has just completed another historical crossing of the country - it loaded up an Audi with enough sensors and computing power to drive itself across the country. From Wired:
An autonomous car just drove across the country.
Nine days after leaving San Francisco, a blue car packed with tech from a company you've probably never heard of rolled into New York City after crossing 15 states and 3,400 miles to make history. The car did 99 percent of the driving on its own, yielding to the carbon-based life form behind the wheel only when it was time to leave the highway and hit city streets.
This amazing feat, by the automotive supplier Delphi, underscores the great leaps this technology has taken in recent years, and just how close it is to becoming a part of our lives. Yes, many regulatory and legislative questions must be answered, and it remains to be seen whether consumers are ready to cede control of their cars, but the hardware is, without doubt, up to the task.
I, for one, am eager to give up most driving. Hell, I don't even want to own a car; instead I want to just summon one when I need it using my cell phone (or eventually via the implants that replace cell phones). The Wired article says that autonomous vehicles will be common by 2040. My bet is that that prognostication is way too conservative.
For more background see my article, The Moral Case for Self-Driving Cars.
Editor's Note: As of February 29, 2024, commenting privileges on reason.com posts are limited to Reason Plus subscribers. Past commenters are grandfathered in for a temporary period. Subscribe here to preserve your ability to comment. Your Reason Plus subscription also gives you an ad-free version of reason.com, along with full access to the digital edition and archives of Reason magazine. We request that comments be civil and on-topic. We do not moderate or assume any responsibility for comments, which are owned by the readers who post them. Comments do not represent the views of reason.com or Reason Foundation. We reserve the right to delete any comment and ban commenters for any reason at any time. Comments may only be edited within 5 minutes of posting. Report abuses.
Please
to post comments
I, for one, am eager to give up most driving.
Never!
"The Wired article says that autonomous vehicles will be common by 2040. My bet is that that prognostication is way too conservative."
Technology wise you're probably right. Cars with the capability for autonomous driving could be common well before 2040 (I think 2030 - 2035 sounds about right)
However I think you may be overestimating the speed that the regulatory state moves at
And completely misreading the way the car market works. This technology will start in the high end range and trickle down from there as costs decrease and technology improves. 2040 seems like a good guess for when the Darts and Sonatas of the world have it as a standard feature.
Adaptive cruise control is already available in luxury sedans. Automated braking systems are already showing in mid-class family movers. There are a couple of vehicles out there that do automated parallel parking.
The car market works somewhat quickly, especially where there are insurance incentives. You can bet there will be an automated driving discount on insurance policies. And there's the possibility that government will force such safety features into all cars by 2022.
And in regards to the regulatory state, it will move slow until suddenly Congress sees a chance for "stimulus" and then you will be incentivized for getting rid of your perfectly good driver-required car. In fact, I can see the regulatory state making driving illegal; they'll do away with driver's licenses and replace it with some massive road use fees.
So the bad news about automated driving is that is essentially going to be a state-mandated ankle bracelet.
I think you all way underestimate how fast tech moves.
2020 roll out for high-end vehicles (limited capabilities)
2025 - 2030 common option for mid-level vehicles and starting to to show up on entry-level.
Ain't nobody selling cars with strong autonomous capabilities to the public until the liability laws get sorted out. Because no automaker can afford to catch the liability for every accident their cars get into.
When that might happen, nobody knows.
I, for one, am eager to give up most driving. Hell, I don't even want to own a car; instead I want to just summon one when I need it using my cell phone (or eventually via the implants that replace cell phones).
Me too, to a certain degree. I actually enjoy a nice drive (my drive from Munich to Neuschwanstein/Hohenschwangau in a manual was an incredibly enjoyable experience), but for nights out or hours/days long highway trips? Yeah, a self-driving car would be great.
But I think we're underestimating the freakout by the politicians and the pigs over this. The car has been a tremendous way for them to undermine constitutional protections, from roadside searches to DUI checkpoints to being able to pull you over for just about any reason. Cars are the way we primarily move about, and they've turned them into another regulated government scheme. Want to drive? Pay for registration. Don't have your papers? Tow and fine and court date.
They expect and rely on a certain amount of revenue from busting drivers for various things. It's part of their fucking budgets. Self-driving cars will crush that revenue. Just crush it. No more DUIs. No more speeding tickets. No more improper lane change. No more "you were driving erratically so I stopped you, can I search the vehicle?" This is going to be a major issue for them. I expect legislation banning these cars all over the place.
And terrorists. And hackers. And don't forget the terrorist hackers with their self-driving cars full of plutonium.
Hackers will actually be a real danger.
OEM's are simply not known for putting *any* thought into securing their devices. From a bluetooth toilet controller that is factory set with a '0000' passcode to the insecurity of USB drives, don't even get me started on 'the internet of things' where your refrigerator can order milk for you.
They will be an enormous danger. These things have to be connected to the internet to navigate and communicate with each other. Once you are plugged in, you can be hacked. There are so many opportunities for crime. Hack someone's car and have it quietly pull out of the driveway late one night and use it to run drugs or commit a crime and then slip it back in before the owner wakes up and let the cops hunt for him and his car. Won't that be great?
These things have to be connected to the internet to navigate and communicate with each other.
No they don't. Airplanes aren't connected to the Internet and do navigation, communications, and surveillance just fine.
Airplanes' on board computers do not talk to each other. That is what pilots do. And there are a lot fewer planes than cars. Maybe these things won't be hooked up but I seriously doubt it. And if they are, they can be hacked. Moreover, even if they are not, if they have a wireless interface, which they will have to have to communicate, they can be hacked. It doesn't matter if they are not hooked up to the internet. If the computer communicates with the outside, a hacker can get into it and take it over.
Assuming the aircraft has an operating transponder they do communicate with each other for proximity info.
Which is probably the way the government will come to track your movements.
In the first decade or so, there won't be enough automated vehicles on the road for anyone automated vehicle to assume others are out there. So these vehicles must assume non-cooperative traffic which is actually worse that modern aviation which has TCAS.
navigation is GPS. local control (avoiding collison) is onboard sensors. internet has nothing to do with driverless cars. how can you be so wrong about EVERYTHING?!
No way are the cops giving up the DUI and ticket racket this easily. For this reason they are not going to allow these cars without a sober driver there to take over in case of failure. That way they can still bust people for driving drunk. In addition, if you have to be there sober and paying attention so you can take over if necessary, you might as well be driving. So that requirement will greatly limit the appeal of these cars beyond nerds like Ron who will buy any new technology.
What we will get is the opposite. We will drive and the car will stand ready to take over if we screw up.
Epi, you aren't cynical enough. Government will find automated driving a cash cow in comparison to your examples, especially since it's all electronic tracking and requires almost no labor like police stings and a court system.
Mind you, they won't actually get rid of any government employees, they'll just get reassigned to a new shakedown scheme. Certainly you must be doing something illegal in your house that they might be able to detect with radar.
Epi,
I am not sure an outright ban would fly...but a myriad of regulations, nigh impossible to meet standards, plaintiff lawyer aggression and such will slow this for some time to come.
The cops/cities/counties would see a quite significant chunk of $ from fines, assessments, and such go away - and, "rolling Probable Cause" as you also mentioned. The G could get its money through "permits" "licenses" and "fees"...but the cops, they would take it in the shorts, hard.
Thank god these checks upon ingenuity are in place to save us from the companies that are so dutifully attempting to kill off their customer base.
Sign me up. Driving is a waste of brain cells that could be otherwise used commenting on H&R. Long overdue.
The car did 99 percent of the driving on its own, yielding to the carbon-based life form behind the wheel only when it was time to leave the highway and hit city streets.
As I said in the last robocar thread, the technology to allow a car to drive autonomously on limited access highways is nearly here. But a fully autonomous car that can drive anywhere, anytime, in any weather will not happen in my lifetime.
wrong
Unless you are going to die in the next 10 years.
And I totally disagree with you. So there. 😉
As someone with a PhD specializing in robotics, I think he is correct.
For deep structural reasons, the current state of tech is quite good for designing robots for very specific tasks. It is not good for designing robots that have generalizable skills. To get there will likely require decades of research and development along completely different theoretical lines.
As someone with a PhD specializing in robotics
Hazel, you've been holding out on us. 😉
Then, how can this be?
Show me a fucking google car driving in an Iowa blizzard on a gravel road.
Not a huge leap in technology.
We completely understand the dynamics involved and we can certainly write code to detect and correct for it. My 07 truck automatically makes a correction if it detects a skid. It's just a matter of incorporating feedback loops to rapidly correct, which is what a human does. I hazard to guess, it won't take very long to make the computer significantly more precise and faster than the human. Same reason the autopilot can fly better than I can.
Skynet started on a road trip, you know.
"I Have No Gas And I Must Scream"
+ 1 Dangerous Vision.
Goodbye to public transportation schemes. All those union jobs, a distant memory.
What a wonderful technology for the disabled. Freedom of movement, autonomy. No more waiting for a government supplied minivan driven by a surly attendant at the taxpayers expense.
Yeah, government will never let this happen.
So goodbye yellow brick road
Where the dogs of society howl
Someone, I think it was 'The Antiplanner'
http://ti.org/antiplanner/
Calculated that it would be cheaper to buy all the transit riders in Phoenix a Prius and pay for their gas for a year than install their light rail network.
Soon, they'll be able to have the advantages of both 'public transport' and private ownership of a vehicle.
How much to convert all the metro/subway systems into roads? We should start now.
Thanks for the link Agge. More good reading. 🙂
Are you kidding? Your driverless car will be property of the state, you merely get to purchase exclusive use rights to it.
How does the Delphi autonomous car drive through tunnels?
GPS even with satellite augmentation is not good enough to keep a car in a lane. Visual sensors must be used to see painted road markings to allow the steering system to stay in any given lane. So driving through is a short term problem that does not directly influence how to keep a car in its own lane.
In fact tunnels are usually the easiest because they tend to not have pedestrians and intersections so you only have to worry about the other idiots on your road.
Limited access highways do not have passengers either. And all traffic is basically going the same way you are. This is why the problem is solvable in the next few years.
fuck. pedestrians.
Although it was a great fuck up.
Right. An interstate highway is a pretty controlled environment. You can only get on and off at certain points. There are no stop signs, speed bumps, or left hand turns. There are consistent lane markers and a uniform smooth surface to drive on.
Except for the "Road Construction Next 10 miles" then they shut down the westbound lanes, funnel all eastbound traffic into the right lane and send the westbound traffic across the median into the left lane of the eastbound side. Very common in the rural stretches of the Interstate system.
Disengage the autopilot; drive by hand for 10 miles; then re-engae the system.
Vehicle operation always in command of the vehicle.
operator ... preview, preview, preview
There are no stop signs, speed bumps, or left hand turns.
Well, there are exits from time to time from the left lane. Not as common as exits from the right lane.
Other than exits, of course, there aren't turns at all on interstates.
I will never give up having a 'non-autonomous' car. But I will certainly appreciate the freedom that having a second self-driving car will provide. If nothing else, bye-bye airplanes except for the most time-sensitive domestic trips.
And my regular car - it will give me an excuse to 'apocalypse-proof' it.
Fury road, here I come!
The car did 99 percent of the driving on its own, yielding to the carbon-based life form behind the wheel only when it was time to leave the highway and hit city streets.
So it's like a modestly improved cruise-control. Yay.
Significantly improved, I'd say. Not even the lane sensing, adaptive cruise control will allow you to kick back and read a book during the trip.
From the Wired article:
The Delphi caravan (the self-driving car, a follow car with more personnel, and a Winnebago full of PR, photo, and video folks) followed a southern route, largely to avoid snow. Apart from the shock of realizing just how long it takes to drive across Texas, the biggest scare of the trip came while crossing a double-decker steel bridge on the drive from Philadelphia to New York. "I saw that bridge coming, and I thought, 'Oh my gosh, this is gonna be a grab the wheel moment,'" says Katherine Winter, a Delphi software engineer. That's because being surrounded by metal plays hell on radar by making it difficult to discern what's a threat and what isn't But Delphi's refined how its software understands the radar data and uses the other sensors to augment it. "It actually outperformed what we thought it would do," Winter says.
The referenced Wired article does not list all the equipment installed in the test vehicle to make the demonstration work nor does it discuss the equipment and installation costs for the demonstration. First demonstration vehicles full of engineering prototypes are routinely 10 times the cost of a production system. But still, it will be a major task to get a fully functional system down to a price point that makes sense in a mid-sized family sedan.
That being said, it is an impressive demonstration.
It is. But with a ton of qualifications. This is impressive but it is still miles away from the kind of thing Ron is drooling over. What people like Ron forget is that economic decisions are made at the margins. It doesn't matter that a technology is better than what we have. It has to be enough of an improvement to justify getting rid of the current technology. And that is not as simple as just being better.
Having a robo car is going to cost more than having a regular one. So it is going to have to be enough of an improvement to justify the cost. Sorry but glorified cruise control isn't going to do it. Again, what you will see is various parts of this technology put on cars piecemeal to improve the safety and driving experience.
As I posted earlier, what Ron wants ain't gonna happen in the next 30 years.
But super-cruise-control long-range highway driving will be here in 5.
Mercedes posted a cool video of a self-driving semi-truck in the last week or so.
I had a loaner 2014 Merc C class a couple of months ago. The back up camera had a digital grid and trajectory for you. It made backing up virtually idiot proof. That kind of shit is already here and is only going to get better. Like I said above, the future is going to be the opposite of what Ron thinks. It won't be humans waiting to step in when the robot fucks up. It will be the robot waiting to step in when the human driver screws up.
No it won't. The robot will be a tool of the government. You will never be ALLOWED to step in unless you have paid for the privilege.
Only if the robot gets good enough. And yeah, Ron is a fucking moron for thinking this is going to be good for freedom.
Yeah, this is what I'm talking about. There are all sorts of situations that are going to fuck with the sensors and autonomous control algorithms. Driving on snow or slush. Passing under a metal bridge that screws up the radar. Driving on an unpainted road where you can't see the lane markers. Rough road surfaces. Unexpected, oddly shaped objects (birds, cattle, pedestrians, traffic cones).
The other thing is that the current systems rely on expensive sensors, largely lidar and radar. Vision-based systems are not there yet, and until they are, this will be too expensive to be in cars that regular jerkoffs drive.
That is a big deal. I can't see vision sensors being here anytime soon. That is an enormous engineering problem.
Don't I know it. I'm cautiously optimistic that I can make some decent cash working on it. We'll see.
It is one of those things that sci fi takes for granted and sounds easy until you start thinking about it. When you think about it, it is a hugely complex and difficult problem. Our brains are a lot more amazing than we often think they are. The ability not just to see but also comprehend depth and distance without reaching out and touching the objects some way with sound or electromagnetic waves is astounding.
9 days on a bullet run? What took them so long? Did the car want to see the sights on the way? Because there's a whole lot of nothing between Cheyenne and Chicago.
? ? ? ? LIFETIME OPPORTUNITY ? ? ? ? ?
My friend's step-sister makes $72 /hour on the computer . She has been unemployed for ten months but last month her pay check was $14639 just working on the computer for a few hours. Pop over to this web-site ??????? http://www.jobsfish.com
No one in the least concerned that your movements are now available for government tracking?
I am.
Nothing about robot cars requires government surveillance. They don't even require internet connectivity or GPS navigation, although that makes the localization problem much easier. And as soon as these become commonplace there will be a hacker culture devoted to disabling any government tracking that they have.
Nothing about robot cars requires government surveillance.
Legislation is all that is needed.
At the very lest there will be two classes of drivers/cars: the government-controlled one which will be relatively inexpensive compared to the super-expensive citizen-controlled one. You'll gladly take the government-controlled one because it's only $150 a year versus the $2000 a year driver-controlled license which is mostly for old car enthusiasts.
This is all good progress, to my mind. Good for Delphi.
I expect I'll always have a meat-driven car. Probably even the ones I own now, as I am not excited about the next gen of mandated tracking technology.
But if I had an autonomous car, I'd use it, probably more than the meat-driven car.
my friend's step-aunt makes $73 hourly on the internet . She has been out of a job for seven months but last month her income was $19815 just working on the internet for a few hours. pop over to this web-site....,
??????? http://www.work-reviews.com