Every year, more than 30,000 people are killed in motor vehicle accidents in the US, and many many more are injured. Humans, in general, aren’t great drivers. Until dependable self-driving cars make their way into garages and driveways across the country, there is still a great amount of work that can be done to improve the safety of automobiles, and the best hope on the horizon is Vehicle to Vehicle communications (V2V). We keep hearing this technology mentioned in news stories, but the underlying technology is almost never discussed. So I decided to take a look at what hardware we can expect in early V2V, and the features you can expect to see when your car begins to build a social network with the others around it.
It All Started a Long Time Ago
The origins of V2V communications date back to 1991, in bill titled the Intermodal Surface Transportation Efficiency Act (ISTEA). ISTEA is best known for providing the impetus for high-speed rail lines that are still unbuilt 20 years later and the requirement for airbags to be installed in all cars and trucks sold after 1998. ISTEA also included a provision for research into something called the Automated Highway System Program (AHS), an effort that would lay the groundwork for V2V communications.
The product of the AHS program was a few miles of I-15 in San Diego being used as a test bed for technologies for the far-flung future. In 1997, the Automated Highway Demonstration commenced, with vehicles equipped with sensors, cameras, throttle, brake, and steering actuators, laser and millimeter wave rangefinders driving along a stretch of highway embedded with magnetic nails. If this program were enacted across the entire United States, it would have been fantastically expensive, both in retrofitting all highways with the required sensors, and adding thousands of dollars to the price of a new car. It was, however, the genesis of Google’s self-driving car concept, and all the autonomous cars that will be taking to the road in the next decade.
A Car Talkin’ Convoy
The AHS Program demonstrated amazing technology, and was the beginning of V2V communication. In this program, individual vehicles could join a ‘road train’, a convoy of cars, each drafting behind each other to reduce wind resistance, drag, and fuel consumption. This is something that was only possible by giving the computer direct control over the steering, gas, and throttle, something the current proposals for a V2V system are not designed for.
https://www.youtube.com/watch?v=Dtj9J1A-PMk
In 2007, the National Highway Traffic Safety Administration (NHTSA) published a framework regarding the future of V2V communications, and last year released a readiness report on V2V technology. This technology isn’t available yet, and even the NHTSA won’t deliver a Notice of Proposed Rulemaking until next year. The system will surely save lives, but it is not as magnificent as self-driving cars would lead you to believe, and surely does not live up to the promise of twenty years ago.
The Wireless Network
According to the NHTSA readiness report, future V2V technologies will communicate over currently unused frequencies in the 5850 to 5925 MHz band. This is just above current bands used for 5GHz WiFi, and will, of course, require new chips, tools, and protocols. Outside of a software defined radio – and a good one at that – you won’t be able to listen in on V2V communications. But if you could, what would you be able to see?
![Exploits, like [Charlie Miller]'s hack of the on-board entertainment system in a Jeep Cherokee presented at DEF CON this year, will hopefully be impossible through V2V communications](https://hackaday.com/wp-content/uploads/2015/08/exploit.png)
But what about controlling cars by feeding them false information. In 1997, a car participating in the AHS Program would transmit information about its speed and upcoming obstacles to cars behind it. Surely under the proposed V2V system, someone could be smart enough to spoof information to a tailgater, sending them off the road, right?
This is not the case, and it’s not for any technical reasons. Auto manufacturers are reluctant to put any actuators on the throttle, brake, and steering mechanism for liability reasons. Car manufacturers don’t want people with SDRs to control most of the cars on the road, it seems. The current recommendation is to only display notifications to drivers, allowing them to react to upcoming conditions. V2V is not a car that drives itself, but it could be used by self-driving cars in the future.
V2V is still a long way off from being a reality – at the earliest, it may show up in cars in the next few years. It’s even further off from the technology demonstrated back in 1997, but self-driving cars are quickly filling that gap. It may not be game-changing, but V2V communication will drastically decrease the number of deaths on US highways, and help in a vast reduction in the number of accidents.
I don’t see V2V being great for control anyways. If you have radar in your front bumper to watch the car in front of you, how many milliseconds would you save (if any) if the car in front of you broadcast that it was stopping suddenly?
I think it could be fairly considerable as the V2V system would be broadcasting when the brakes are applied. There is a measurable delay between initial application of the pedal/hydraulics and the brakes actually taking effect. Using V2V would give the vehicle behind a heads up over just relying on radar.
There would also be delay in a radar system as it has to take multiple measurements over a period of time to determine if the car is actually braking or if it’s just a variance in the measured distance as well as determine the rate of braking. In comparison all of this data could be relayed in a single V2V broadcast.
Not only that, but the v2v messaging also reaches the 6th truck in the convoy at the same time as the others, and they can all brake together. With the radar, the delays just add up for each vehicle in the convoy.
But it also opens a huge can of worms because as soon as you start accepting data from outside your own system, you have to worry about “attacks”. Is the benefit really worth the risk?
+1
I still think that following proper road precautions are better rather than having the driver relying on a system like this. I would like to have this system to only alarm me if I’m dangerously close to another vehicle at a certain speed, but for it to automatically engage the brakes for me is out of the question.
I agree that it should not be used for control. Instead it should be solely informative.
Building some sort of mechanical device or performing a CAN bus hack would start a chain event going to all the cars behind you with unknown consequences (or intended if you are feeling malicious).
I consider the cost of a GPS as minute compared to the functionality you gain and think that it should be incorporated into all vehicles. Add V2V and you will not only have traffic reporting for the GPS solved, but also, you can start adding the vehicles onto the map if each vehicle transmits it’s length and distance to the car in-front. Breaking vehicles are shown in red and so on.
A better implementation would be to incorporate it into mobile GPS devices that stick to your windshield. Simple and non invasive.
When driving around the countryside where I live (rural: hedges, tractors, horses, cyclists, walkers, etc) I’ve thought about the merits of V2V type comms as a sort of hybrid radar system where rather than looking for an echo you gather the data from vehicles heading in the other direction or a bit further up the road. Rather than be invasive and, say, automatically apply brakes, I’m thinking of a slightly more passive system that simply provides the driver with additional notes about the road ahead (like signs that road ahead is twisty) but in a more dynamic manner.
Current options are either: end up crawling around every blind corner paranoid about what is there or be quick to leap on the brakes. Having, effectively a potential set of eyes around a corner, it shouldn’t stop one being cautious but should reduce some surprises (aka ‘accidents’) which is surely the name of the game? (ie can’t stop everything but at least improve the odds). That and knowing when roads are blocked ahead (for whatever reason) so having a forewarning of the need to take a detour without needing to wait for the occasional traffic report on the radio (which only really cover the major roads).
Oh, also being able to send messages to drivers who are unaware that their bonnet/fuel-cap/boot/door is open or there is some trim/shrubbery/road-kill hanging off would be quite handy (waving and flashing lights can only get across so much information)
Better if your own car can tell you fuel cap/ door etc is open rather than rely on others, most cars already do
Cars already tell you if your indicators are going but it still doesn’t stop people leaving them on :-/
(I suppose if people can ignore one thing then they are just as likely to ignore another)
“Until dependable self-driving cars make their way into garages and driveways across the country”
Really? I’ve been constantly surprised to see blogs in the last two or three years about self-driving cars on public roads. It seems like poor and risky engineering to me. Who would be guilty of homicide if a “self-driving car” caused a fatal accident due to a software bug or an erroneous action in an unpredictable situation? Would the software developers go to prison for that?
Wouldn’t it make far greater sense for the billions of dollars to be invested in separate roads or tracks where such vehicles can drive? Why mix old tech (tar roads – a.k.a asphalt in some parts of the world, painted lane lines, human readable stop and yield signs, traffic lights, traffic circles, intersections etc. etc.) with new self-driving tech? Something similar to the rail network would make far better sense, where “road-signs” more appropriate to computer driven vehicles are used – RFID, V2V etc etc.instead of white painted lines on the road, which are much more difficult for a computer to decipher. What about some kind of electronic lines in the road emitting RF (including traffic info) at a standard frequency? This could be detected by the vehicle to see if it in on track, rather than the much more complicated system of having stereoscopic cameras with optical recognition software trying valiantly to determine the vehicle’s position relative to the lines. Or if it’s gonna kill a child on the pavement (a.k.a. sidewalk) and send the developer to prison.
Come on!!!
Because building an entirely separate transportation network ranges from impractically expensive to impossible, depending on location. Under a separated system there wouldn’t be last-mile delivery of people outside particularly dense urban areas.
Fundamentally, white lines on pavement are very, very cheap and easy to maintain compared to RFID or active systems.
Don’t be like that, our High-Speed Rail To Nowhere is awesome!
“Who would be guilty of homicide if a “self-driving car” caused a fatal accident due to a software bug or an erroneous action in an unpredictable situation?”
Nobody. I mean, as a driver the worst you could get is unintentional manslaughter, and you’d have to be driving in a dangerous manner or traffic laws.
“Would the software developers go to prison for that?”
Only if you could prove gross negligence, violation of certification procedures, or intentional sabotage. Event then the liability would probably fall on the company, not the employees. I’m sure you could find precedence for that with current vehicles, they may not be automated but there are a lot of systems that have to work correctly to make you safe.
Remember, automated vehicles will crash. They will cause fatalities. But as long as they statistically cause less fatalities than humans, they are still a win. And lets be frank, most human drivers are not that hard beat.
Laws vary tremendously from one part of the world to another. Even the criminal justice codes vary tremendously from one part of the US to another.
That having been said, I can easily imagine an ambitious, grandstanding, political District Attorney charging the driving software dev team collectively with what US law calls “manslaughter,” which the law books generally describe as “unintentional death of a human being resulting from depraved indifference to human life,” the sort of thing that was intended to punish people who drop concrete blocks from bridges into heavy traffic speeding by below, or people who kill someone while driving under the influence of alcohol or drugs.
There is also the civil law in the US, which is wholly different from the criminal law and which is tried in a separate court system. Americans have a nasty reputation the world around for being litigious. If someone put a self-driving car on the road, and this self-driving car was involved even peripherally in an accident in which someone suffered great bodily harm–it wouldn’t even have to result in a death–the injured person and his relatives would suddenly find themselves surrounded by civil lawyers parachuting from the sky and popping out of the cracks in the sidewalk, tripping over their own feet to offer their services in the inevitable billion dollar lawsuit.
This, as much as the current state of the technology, is what keeps self-driving vehicles off the roadways, at least in the US. If any of them ever injures someone–and statistically it is inevitable, even if they turn out to be vastly more skilled than human drivers–the manufacturer and designers are guaranteed to get sued down to an inch from the ground.
Wouldn’t it make far greater sense for us to have separate infrastructure for every transportation type. A road each for cars, true-autos (my new trademark for autonomous vehicles), pedestrians, motorcycles, bicycles, joggers, skateboards, rollerblades, horses, segways, hobos, parades, demonstrations, even a hot spare just for road work.
In all seriousness, all of the marvelous individual transportation solutions mankind has devised has had it’s share of people who didn’t think intermingling was safe, and in every case I am aware of so far we have settled on intermingling either in part, or commonly in whole. For the first time we don’t have an animal making the dangerous decisions, I am willing to wager that is the best improvement we will get anytime soon.
Machines being better than the humans at driving on a statistical scale is almost a guarantee (given sufficient sensor, actuator, and processing capabilities). Auto accidents represent one of the greatest dangers in the modern world, an accident caused by a machine error would be sad, but it would be far less common, if only for the liability reasons you bring up. Just about any given machine issue can be addressed, human error can’t be in any general case.
All the above isn’t to say that I welcome losing my control, but I understand many reasons why it serves everyone to move in that direction. I am more worried about police powers or assassination by autonomous vehicle than I am about the more obvious and direct risks. I am a controlaholic, I prefer the least automated machine possible for almost any given job.
I have never seen a decent comparison of traffic fatalities for equine vs human error, I wonder what that would show…
A separate problem is that any such automated vehicle system is almost certain to be under government control, complete with centralized government kill-switch, which is certain to be abused by governments that wish to limit their citizens’ freedom of movement.
We already have such at network in many places – it is called a railroad.
What we should do is move much of the transportation to it, and off of the highways.
Much of interstate trucking could be done much more efficiently using the rail network.
I can see the update now…
Car443: Doing 40mph over the speed limit, out of my way!
Car447: slowing down rapidly because I was texting and did not see the semi truck ahead of me.
Car621: weaving into the rumble strips again, driver is watching netflix on his phone.
I worked for a major car manufacturer once-they resisted gps as long as they could out of fear of lawsuits if someone got sent to the wrong address. Self driving cars would be a heyday for lawyers.
Self-drivers are all over Mountain View this summer. Google has two people inside each car, maybe one is legal counsel.
No. The second is a blow-up doll so they can use the HOV lane.
The first company to sell an automated car will likely end up getting sued into bankruptcy and the major automakers know this.
V2V will bring on interesting new opportunities as well. E.g. imagine some script kiddies broadcasting brake signals from a laptop next to the highway…
> 5MHz WiFi
Where could I find some of this 5MHz WiFi? Im stuck with my lame 5GHz WiFi
SoCs with radios are dirt cheap now a days that, compared to a lot of the costs associated with road maintenance, deploying them on a large scale is almost negligible. Beacons or what have you could easily be added to road signs, traffic lights and mile markers, which in turn would benefit autonomous vehicle systems. As for the cars themselves, we’re already well into drive-by-wire. There’s electric power steering, electronic throttle and ABS systems which can apply braking to individual wheels for traction control. Computing is cheap, cameras are cheap, GPS is cheap, and if deployed by the millions, presumably depth sensing systems like LIDAR would be as well. I think all the pieces already exist to make the system a reality outside of further software development and beacon deployment (not entirely necessary, but I think it would be a good idea).
nonetheless, the tech is really fascinating, and hilariously expensive to implement. If you don’t want to drive take the light rail peoples!
Laying new rail lines near my house would be even more ridiculously expensive. I only drive to work because public transport is more expensive and significantly less convenient. I live in one of the 10 most populous cities in the UK, what do you suggest my parents who live in the middle of the countryside do?
Google posts accident data monthly now about their autonomous cars. The findings are quite amazing- as of June this year their system had not been the cause of an accident, but they have been rear ended a number of times.
I have a really hard time wrapping my head around the legalities of autonomous vehicles. As a pilot, it’s burnt into my head that regardless of the state of the plane I am the pilot in command and am therefore responsible for the safety of the vehicle, my passengers, and everybody else I fly around or over. The grand dream of getting on the freeway, taking a nap, and waking up at your offices parking lot does not seem like something we are even beginning to be close to in the near future, if for no other reason than as a society we are not ready for driver less vehicles.
If you read the detailed description it becomes clear the google cars often enough cause accidents by driving like old people, when they cause accidents that is.
But I don’t like the concept, think about the future, every move you make will be reported to google, and they know your destination in advance. The roads and driveways around your house will be filled with cars with 360 degrees cameras and other sensors, and it will all be networked and accessible by spooks of course.
And every person that gets into your car or near your car can be logged.
So yeah, fuck that nightmarish horror I say, it’s not a world that looks too rosy to me.
And as for preventing accidents, I’m sure that if they ship us all to guantanamo or siberia and put us in cages we would also be safe from traffic accidents, but do we want that? Perhaps you do, but I do not.
Not sure what you mean by “driving like an old person”. Some of the Googlemobiles have been rear-ended, some have been sideswiped, and one got into a bit of a squabble with another brand of self-driving car about who got the lane (the one with right-of-way won, and the other did a panic-slowdown.)
The only real problem I’ve heard of was the Googlemobile that got confused by a biker on a fixie, who was doing the back-and-forth standing thing at an intersection.
I mean it’s overly careful and doing the overly slow moves and full stops that normal drivers don’t do to the extend that a computer does, except for some nervous old people who also can have a tendency to do things overly careful causing actual confusion.
And I’m not simply claiming that, it’s described in the google pdf.when they described the various incidents.
And they can of course then say it was the error of the human, but is that completely honest?
And what if it happens on the highway btw, what if everybody is driving a bit over the max speed (as happens in reality) on a stretch of highway but the google car steadfastly stays 5 mph under it to maintain a legal speed? At that point a human adapts to the other traffic but then suddenly come upon a robot that does not. Now legally the human that might hit it would be wrong, but in reality I would put some blame on the robot too.
If the software is setup so it drives like an old person it may cause accidents when people pass it.
Police forces will mandate it has some sorta remote shut off and this will of course be abused by law enforcement and will be hacked.
If an average traffic cop can make one car pull over a couple of bored teenagers with modest hacking ability will be able to put together a box that could make all the cars on a freeway off ramp stop and pull over for laughs.
Even if it has no internet connection which would solve most hack attempts it still may be possible to crash them with a high power IR laser.
It would be nice if it could be done without invading privacy and tracking you and controlling you.. but of course they’ll never implement things in that way.
It could be implemented without tracking and invasion of privacy as bad and ironically this would make it a lot safer as the same back doors that allow government spying also enable hackers.
But there are shady entities drooling over the possibility being able to watch everyone’s movement.
Even greedy municipalities would love the ability to issue speeding tickets through the mail without the need for expensive cops and speed cameras.
“Auto manufacturers are reluctant to put any actuators on the throttle, brake, and steering mechanism for liability reasons.”
Really? Throttles have had automatic actuators for years. They’re called cruise control.
My new, cheap, base-model Subaru has computer-mediated actuators on all three. The throttle is controlled by computer, with guidance from the angle sensor on the throttle pedal. The brakes are controlled by the ABS and vehicle stability system and can either apply brakes or release them as conditions require. The steering, although connected to the steering wheel, is directly controlled by an electric motor (not hydraulic power assist).
The Subaru Eyesight option (and similar ones offered by many manufacturers) will automatically apply the brakes to avoid a collision. (It will also monitor lane position and nearby obstacles and warn the driver, but not actively avoid striking a cyclist, for example.)
Yes, the Naders have repressed and delayed useful progress, but it’s still happening.
There is no way in hell I’d buy a car with a camera systems that automatically applies the brakes.
Now a Doppler radar and an audible warning I could tolerate having present.
Or, we could take the train.
Cause laying tracks is the cheaper option.
Well yes it always is, it’s all about planning the city.
> reluctant to put any actuators on the throttle, brake
they do. stability control, traction control, and abs are mandatory.
If the internet and its 87.5% operational availability is any benchmark for this technology then all I can say is……we are a long way off.
How do V2V systems work and still provide privacy?
One of the desiderata listed on the national highway transportation safety proposal on V2V was to allow the blacklisting of devices.
(Another was protection of privacy).
These goals seem to be mutually exclusive (or at least very difficult to provide one without making tracking easy).
So how is this done? (If somebody has been really clever and worked out how to do both at once, I am curious how).
Or is this just wishful thinking/lip service on their part.
If this is just an information system – what evidence is there that the additional information increases safety?
Additional information can easily overwhelm the operator. It can also make the operator less cautious.
Will the system be required to be experimentally validated to actually present information in a way that improves safety? (This should be mandated.)
Or is this just another case of people pushing a “solution” that looks like it may help, but could be useless, or even harmful.
(e.g. there is some evidence that calorie counts on menus at fast food may not reduce over-consumption of calories.
Or the assumption that activities that “should” improve self esteem will improve scholastic performance (just because people with high self esteem do better in school).).