Vehicle-to-Vehicle Communication Could Replace Traffic Lights, Shorten Commutes
spectrum.ieee.org
spectrum.ieee.org
I have yet to see these sorts of considerations in any V2V communication system. The idea that my car might act on (or propagate) incorrect/sabotaged information that it receives from a "peer" is a terrifying form of fragility.
This sort of failure mode needs to be studied and addressed directly before any of these systems are deployed.
It should also be assumed that these new patterns of information propagation will create a huge financial incentive for people to sell after market modifications that exploit the trust models of these protocols.
Even if total fatalities drop, if the reason for the remaining deaths is that your car will arbitrarily lie with no justification, then you’ll lose trust in the whole thing.
Systems that require trust should be reasonably perfect, so as to maintain that trust. Otherwise you’re really only going to get away with it by forcing it down the consumer’s throat, by top-down approaches (gov regulation, contracts with the ceo, etc).
And when you’re doing that, who cares what the failure scenario and rate is? Lies to you 20% of the time, and its still on your head. 30%? 80%? The only group that needs to trust it at this point is management; they’re n steps removed from the issue, so as long as you can keep them from looking too hard, you can go as awful and shitty as you want.
Trust takes years to build, seconds to break, and forever to repair. Or you just take it to management.
Your website doesn’t need much trust, and your text editor needs some but not much. But your car? It most certainly does
My point is that V2V systems and research papers I've seen just don't make meaningful claims about safety. They instead make claims about convenience and efficiency, which are not substitutes for safety.
We know to test vehicles for crash safety before putting them on the road.
While I believe there are certainly ways to make use of V2V communication that increase overall safety, I haven't seen anything remotely resembling a crash test for V2V systems.
Creating a system that relies on the correct behavior of all components is not a recipe for safety or reliability. This is especially true as the number of components increases (this happens when a car exchanges messages with all the cars around it).
We need V2V systems that rely only on correct behavior of any component and allow for malicious behavior of some components.
The idea that a protocol version mismatch from some rolling deploy can cause injuries in cars made by other manufacturers is the sort of thing that I haven't seen a single person point out. How would something like this even be caught and debugged?
Advocating for an ecosystem where these things are likely but neither addressed nor considered is just plain irresponsible.
The level of security and approach to security will largely correlate to the types of messages that such a system ultimately needs. This sounds obvious, but I think it really isn't. In a world of self-driving vehicles with lidar + optical sensors, how much v2v communication is required beyond the sensor data?
Until that question is answered, it might not make sense to primarily focus on the security and reliability of said data.
I hope you're right and that once we establish best case utility, as a community we refocus on handling component failures more gracefully.
Although it's unclear if the second and third order system effects in our financial system will ever get that sort of treatment. So let's hope driving gets a bit closer to flying (and farther from wall street) in terms of attitude towards safety.
EDIT: clarity
How do you figure?
The arbitration system is designed to let chatty-but-low-priority messages try to talk as often as they'd like, but only succeed when higher-priority messages allow it.
Short of faults that actually electrically disable the bus, it's pretty good.
From https://knowledge.ni.com/KnowledgeArticleDetails?id=kA00Z000...
I'd rather just have to trust the safety standards of the manufacturer of my car (and to a lesser extent the cars I might directly collide with), not the safety standards of every vehicle within transmit distance.
Especially the associated threat modelling and engineering principles!
So maybe car manufacturers will first incorporate a driver assist that tells you if that car is stopping in front of you. If that fails the driver is still in full control.
Everything I have ever seen about vulnerabilities in car software systems has indicated a very poor understanding of the threat landscape on the part of car manufacturers and an embarrassingly weak ability to competently deal with these threats. So far this has been understandable since cars have been minimally networked, but going forward, I agree with GP that the appropriate term is "terrifying".
But if we can be certain based on the physics of the system that we are talking to the car in front of us, the fact is, there’s plenty of ways you can commit vehicular homicide today, and V2V doesn’t seem like a particularly “worse” way, and frankly, one of the most traceable ways you could probably try to hurt someone.
So while certainly you need to defend against broken and malfunctioning input, I’m not quite convinced the malicious input is actually a case that needs to be specifically defended against.
The vehicle will have a “flight envelope” based on its own local sensors and rules just like modern aircraft that don’t allow even bad inputs to stall the plane. The inputs from V2V would not let you leave the envelope any more than the autopilot inputs would. I believe the steering wheel would still be allowed to exceed the envelope, for as long as there is a steering wheel.
However, intentionally causing injury isn't something that should be ruled out, either. It's only "traceable" if someone is sending the signal from their own car, registered under their real name. If someone pulls the transmitter from a junked car (or build their own, etc.), they could e.g. conceal it near an intersection, wait a day or two, and trigger it remotely, or attach it to the underside of someone else's vehicle, etc.
Someone could also jam the signals to potentially cause everything to stop working.
I'm with the crowd that thinks this is an inherently bad idea. The data is entirely untrusted, which makes it essentially useless for determining anything other than "there seems to be a radio transmitter at a particular location", and that's only if there are enough sensors to triangulate signal sources accurately.
[1]"Likely" means how often that function is used. As an example, applying brakes is extremely common. So any electronics/software involving brakes is considered very likely.
Now whether companies like Tesla actively follow the ISO - I don't know.
It's also the part of V2V that seems to be most powerful and least talked about.
[1] https://www.wired.com/2015/07/hackers-remotely-kill-jeep-hig...
V2V communication means BT exploits can become worms even more easily, spreading from phone to vehicle to vehicle.
Turn signals also get misused and misinterpreted but no one suggests that means we must go without.
That is, as a human I know that the drivers around me might have broken lights or misuse their signals.
I also assume that other drivers do not have my best interest at heart.
A reasonable V2V system should be expected to handle these scenarios just as easily as they handle wireless interference and packet loss.
That said, if you want to lose sleep at night, I suggest you look into the strict authentication and jamming resistance of ADS-B (effectively the aircraft equivalent of such a network).
There are many intersections where the car would need to communicate with the crossing car and the intersecting road is behind a hill, trees, or building.
Yes, attacks are an enormous concern, but comms and hardware failure are also very significant & non-trivial to solve.
What you mentioned probably is going to be high-order technical problems.
I guess these people never heard of common techniques to detect vehicles - in surface wire loops to detect inductance changes from a car or bicycle, and cameras. When the detectors are far enough back from the intersections, you have enough time to (safely) cycle the signal for lone cars in the middle of the night, without the car having to stop -- my experience with well tuned traffic controls was i would just need to take my foot off the accelerator, and the light would usually change to green before i would have started braking.
That works well in a grid based suburb, I understand Pittsburgh has a lot more constraints, but detectors are still useful, even if they're only at the limit line.
Reading further:
> It’s important to note that VTL technology needs no camera, radar, or lidar.
Yes, trusting the communications network, without verifying the situation sounds like a great idea. Just like I always hit the accelerator right when the light turns green, without regard to if the intersection is clear, or if there are any signs that any other vehicles are likely to pass through soon.
Sometimes, I feel like people working in the self-driving space, and the maps for driving space have never actually driven a vehicle on the road.
Here in Thailand, they definitely don't have detection wiring in the road surface, and lights are definitely just on timers.
The difference here is, a) a good chunk of people ignore red lights anyway, and b) at night or other apparently less busy periods, they just turn the traffic lights off and it's a free-for-all of chicken.
Nothing that relies on the driver following an indicator (i.e. an in-car red light) will work here - most people would just put tape over it because it's annoying.
More importantly, have never driven a _variety_ of vehicles down the road. I ride a motorcycle, it's clear most of these plans have no concept of what that's like.
Motorcycles work quite differently than cars in many, unintuitive to car-driver, ways. Until I began riding, I thought I understood, but now my POV is forever changed in a great many traffic situations.
This proposes to replace/augment traffic lights with radio transceivers and in-vehicle displays. What's different about how motorcyclists interact with traffic lights? Isn't it just stop on red and go on green like for any other vehicle?
This system uses a transceiver in the vehicle that broadcasts its location. So in the case where a motorcycle has the transceiver, it is superior because you no longer are dealing with a sensor that performs poorly on your vehicle type.
In the case where a motorcycle doesn't have the transceiver, you'd still see a red light just like they do now. And you'd still have to visually check that nobody is coming before proceeding, so that seems identical.
Roundabouts work better to keep traffic flowing in all directions, without complicated signaling of any kind.
They're great in certain use cases, but not city traffic lights
I agree that these are both edge cases and vehicle drivers need to be away of the possible other users. However, as we all know from experience, there are a lot of inattentive (for whatever reason) drivers on the roads. I think roundabouts have their place and can be a great solution in the right situation. But like any intersection design, they are not a silver bullet for all problems.
[EDIT] To give some context to my response, I am a traffic engineer in a semi-rural part of the US. In my area we have around 12 roundabouts of various sizes and configurations serving a large range of traffic volumes and user mixes.
Experience as a pedestrian and a cyclist, I don't drive:
1. Obscured sightlines over the roundabout make timing my crossing more difficult
2. Exit speeds are usually higher
3. Drivers on the roundabout seem to be more distracted than at junctions, since they're also watching the static traffic at each entrance they pass. They seem less focused on what's in the exit path than at muttering 'stay there' to the drivers trying to get on
4. Difficulty of integrating traffic lights for pedestrian crossings
5. Some drivers are just poor or lazy at navigating roundabouts and I can't trust their indicators, I have to wait to see their actual physical actions before I commit to crossing
Crossing a road when cars are not held by traffic lights always 'suck' for pedestrians.
Conversely, most major roads these days are designed to give green waves if you're traveling at the intended speed; you generally only catch the wrong light once. (If that's not the case, removing a pedestrian cycle would not really give your road any more significant priority than it already has, since your road is already being deprioritized for the crossroad.)
a) I walk up, the light is red in my direction; I push the button, and the other direction turns yellow within 10 seconds (often immediately), and shortly I get a walk sign
b) I walk up, the light is green in my direction, I push the button, and get a walk sign immediately
c) I walk up, the light is green in my direction, I push the button, and don't get a walk sign -- but soon the light turns yellow, and the other direction gets to go, then i get a walk sign.
It's not too far off in the south bay (SF bay area).
It's different in the middle of a city like Seattle, where there's always a lot of traffic, so the lights are changing frequently anyway, a lot of the lights just have the pedestrian cycle enabled almost always; and they often won't add the walk sign in the middle of a cycle, when it's already decided to change the cycle in less time than is needed for a pedestrian to cross.
They explicitly addressed exactly those two things in the article:
"One early idea was to place a magnetic coil under the asphalt surface of a road to detect the approach of vehicles along a single route to an intersection and then adjust the duration of the green and red phases accordingly. Similarly, cameras placed at intersections can be used to count the vehicles in each approach and compute how best to time the lights at an intersection. But both technologies are expensive to install and maintain and therefore only a few intersections have been fitted out with them."
Huh? This is already common practice, why is this "an idea"?
Seems like they'd also get more data (number of cars, individual positions and headings) and may be able to do something smarter with additional data.
It's also not clear to me that adding a module to every car, plus adding modules for the traffic controllers at every intersection is less expensive than installing modules and detectors at every intersection for existing methods.
Here some traffic lights have cameras that detect when you cross at red and automatically send you a ticket, and it's a grave offense (cost 4 points out of 12 on the license).
This is what is described as a "beg button" by urbanists. Requiring pedestrians to press a button to give them a safe window in which to cross doesn't account for situations where pedestrians may not be able to request to cross due to construction, disability, or other factors.
https://usa.streetsblog.org/2017/03/24/seattle-campaign-to-g...
A far better system for connected vehicles would be pedestrian signals that are always green for pedestrians by default, which will change to permit vehicles to pass when vehicles are present. This is very expensive though. Far better and cheaper to just ban cars.
Drivers can see the walk signal too. It serves as an extra reminder to look for a pedestrian. If it's on every time, then it conveys no information about whether there's a pedestrian this time.
For example, I'm standing on the northwest corner, facing east and waiting to cross. A car is also facing east and will be making a left turn across my path. Ideally every driver looks all around, so they'll see me 90 degrees off to their left, but in practice drivers tend to look in the direction they're going to be driving. This means their field of vision includes the walk signal but not me.
For that matter, I wouldn't hate the idea of expanding the button's functionality to advertise my presence even more. Like sticking a flashing "PEDESTRIANS PRESENT" light right next to the red/yellow/green light.
It seems the safer solution is to let the cars still vote but keep the traffic lights and let them have the final say over who gets the green.
Traffic lights have fail-safes designed to make it impossible for both directions getting the green at the same time, an autonomous voting system among cars has fewer assurances of avoiding that state, even moreso if there's a malicious hacker in the mix that set up a bogus transponder at the intersection.
Why not have smart traffic lights as a starting point? Realtime traffic patterns using google maps and cameras with sensors. Data is stored for historical patterns and for new models to be tested against.
For example, McCain (a popular manufacturer of traffic controllers), has their own TMS (Traffic Management System):
https://www.mccain-inc.com/images/mccain-files/products/cut-...
What I don't think there is yet is any sort of neural network training environment for a TMS, but I might be wrong.
I really hope that vehicle to vehicle and vehicle to infrastructure communication is architected to preserve privacy. I really don't want my car constantly transmitting its license plate number, VIN, or even a randomly-generated GUID all the time to all comers. If it does, it'll end up being like having a license plate reader at every intersection and on every police car: your movements would no longer have any privacy.
However, it's probably a super-hard engineering problem to build a privacy-preserving vehicle communication system, so it's unlikely it'll be done.
> The principle behind the traffic light has hardly changed since the device was invented in 1912 and deployed in Salt Lake City, and two years later, in Cleveland. It works on a timer-based approach, which is why you sometimes find yourself sitting behind a red light at an intersection when there are no other cars in sight. The timing can be adjusted to match traffic patterns at different points in the commuting cycle, but that is about all the fine-tuning you can do, and it’s not much. As a result, a lot of people waste a lot of time. Every day.
Isn't that false? Aren't a lot of modern traffic lights equipped with induction loops in the road surface [1], so they can detect if cars are waiting?
[1] https://en.wikipedia.org/wiki/Induction_loop#Vehicle_detecti...
Correct, but pervasive, automated surveillance and tracking has quite a different quality than requiring identification that's always displayed but rarely read. Just because some new practice can be analogized to an old one doesn't mean it's the same.
As a pedestrian (or, when I was younger, a cyclist), I relied on traffic lights as much as (if not more than) pedestrian crossings to tell me when it was safe to proceed. Removing lights and signage may seem like a "simplifying" or "efficiency" measure to vehicle manufacturers, but it's creating a hostile and/or dangerous environment for other classes of road user ... not all of whom can afford, or can use, some variety of AR system to supplement their merely human senses.
Yes, there's a role for V2V signaling: but as a supplement to—not a replacement for—existing traffic control mechanisms.
The second issue, 100% of vehicles on the system, is more insurmountable. How many vehicles do you see on the road that are older than 5 years? 10 years? 20 years? Quite a few. Many of the people driving those cars fall into two categories, those that can't afford a newer car or collectors. Those that can't afford a newer car likely cannot afford to retrofit their existing car with V2V hardware. The second group, collectors, likely don't want to install aftermarket electronics into their vehicles. The presence of cars without the necessary equipment may result in very severe consequences for the system.
My opinion on improving traffic flow through a network is that improvements to a centralized monitoring system will work better and be more resilient to edge cases. With a centralized system, it is easier to ensure that it is operating as it was designed. If you have tens of thousands of individual systems all trying to operate in concert, it could only take a few to cause havoc. Bad actors and poorly operating units will happen.
But client-server is so much simpler operationally with respect to NAT traversal and CGNAT and IPv4/IPv6 and latency and other real-world networking and compatibility concerns. So I expect a commercial enterprise building "smart" or "self-driving" cars will instead just go for the tried-and-true thing of cellular modems that phone home to a central server. (Or, SSH servers exposed to the Internet, I guess, like Tesla did: https://twitter.com/atomicthumbs/status/1032939617404645376) And if the server shuts down, then the devices become useless (like with so many other smart devices).
If you want to get rid of traffic lights, what you want is a dedicated right of way. It can have train tracks, or it can have asphalt (like transitways in Mississauga or Ottawa, or conventional freeways, highways, or tollways).
Imagine one car approaches an intersection and there are no other cars anywhere near. It gives itself a green and goes through the intersection without stopping.
Now imagine the same situation except there are two cars headed toward the same intersection at the same time, but neither one picks up the signal of the other. They both believe they are the only car headed toward the intersection, they both give themselves a green, and then they crash.
It seems like you either have to add some centralized infrastructure or accept that you are only safe to the extent that your radios are reliable.
Won't you need to have the same assumption if you had a centralized infrastructure? Otherwise how would your car talk to the centralized system?
https://www.arrow.com/en/products/adalm-pluto/analog-devices
Take it from 300MHz-3.8GHz to 70MHz-6GHz
https://www.rtl-sdr.com/adalm-pluto-sdr-hack-tune-70-mhz-to-...
Buy some antennas for the freq's you want to RX/TX.
Sniff or investigate the protocol.
Cause wrecks and brake-checks and other bad things with cars.
I guarantee this will be as bad as ADS-B : https://www.youtube.com/watch?v=CXv1j3GbgLk
We need to find a way to make car-optional transit profitable or socially desired.
> The concept of incomplete penetration of DSRC transceivers brings up one of the biggest potential obstacles to adoption of our VTL technology. Could it still work even if only a certain percentage of vehicles is equipped with DSRC? The answer is yes, provided that governments equip existing traffic signals with DSRC technology.
There is no way you can modify all cars on road to make them DSRC compliant any time soon. OTOH making the traffic lights understand the system is very practical and incremental way of improving the situation.
And one would guess the minute this billion dollar investment works, it will suffer the By-pass paradox and streets are back to the usual level.
(its been a week of traffic accidents and long long tailbacks in London, so i am feeling bitter :-)
The system would require trusted communication.
The system would also have to cope with any number of vehicles not having this feature or with broken/defective feature...
What's the fail safe solution? To rely on traffic lights and priority markings.
We could see small trains of autodrive vehicles tailgating each other on highways, but not huge trains of random participants. The incremental advantage is far outweighed by the associated risks.
What will acceptable de-confliction space be? Ten seconds? Five? One? If you are a passenger in a car approaching an intersection, exactly how close do you want to be to the truck about to t-bone you?
Is this accounted for? I’m envisioning roaming packs of nerds with the digital equivalent of the mythical Chrome Box...
For example, a vehicle equipped with this and with an intention to turn right at a particular crossing could communicate that intention to that crossing ahead of time to the controller that manages the traffic lights at the crossing and communicate an ETA or even negotiate a time slot. The crossing could than take this into account and schedule the light to be green for the vehicle when it arrives or tell it to slow down slightly to match the next possible green light to minimize congestion around the crossing.
This would allow infrastructure to start taking intelligent decisions based on the traffic that is communicating to it. It would provide immediate incentives for car owners to make sure their car is capable of participating. Even cars that aren't participating could benefit indirectly from other cars reporting their observations of other traffic (e.g. Tesla does a lot of object tracking) or by simply following cars that are equipped.
You could add priorities to this as well where e.g. ambulances or other emergency vehicles never encounter red lights and cars ahead are directed to make room. You could also allow people to pay for upgrading their priority so that they encounter less red lights and can get more reliable and faster ETA to their destination.
I’m trying to think about how V2V systems could improve traffic without introducing a huge element of risk. As I think about drivers, they make “offensive” and “defensive” moves all the time. An example of an offensive move is to pull into an intersection to cross through it. An example of a defensive move is to stop or slow your car when you see the path in front of you is obstructed.
Humans convey their offensive intentions in cars all the time, to help others execute defensive responses. When changing lanes, we can use our signal. When two cars arrive at an intersection with a stop sign at the same time, one car might slowly nudge their car forward (offensively), indicating to the other car to wait (defensively).
Perhaps cars will broadcast “offensive” information to allow other cars to respond “defensively” to improve traffic, but the lack of these offensive messages (or even in the event of an incorrect or rogue messages), safety is not compromised (although traffic efficiency would be diminished).
For example, if cars were traveling near to each other in adjacent lanes and car A communicated its intention to merge into the lane occupied by car B, car B could (defensively) make a small change in speed or change lanes to make room to accommodate car A’s (offensive) attempt to change lanes.
In the event that car A broadcasts an “offensive” message incorrectly (or fraudulently), the defensive response of car B does not jeopardize safety (although the rogue message would reduce overall traffic efficiency). If car A does not broadcast the offensive “lane change” message at all, car B should be designed to prevent collision (if necessary, applying the brakes sharply, or executing a quick maneuver to prevent collision). Traffic efficiency would be reduced, but hopefully safety would not.
It is near impossible to control all vehicles for the introduction of malicious actors.
Also, the hardest part won't be the actual communication, it will be the standards that have to implemented through all manufactures who don't want to agree and potentially give up their advantages.
But one really interesting application would be road-trains for freeways. You'd cut down on drag significantly, and could probably make it work on manual-driving roads with object/hazard communication and real-time agreement on emergency maneuvers (e.g. every other vehicle swings out to the breakdown lane, simultaneous braking, etc.).
Drag is ~2/3^rds of the energy loss of a big-rig on the highway, so imagine the kind of range a battery electric truck could get in 'road train' mode.
[0] https://www.reddit.com/r/gifs/comments/5dervt/what_a_driverl...
There is an intermediate "rich" communication light-set that speaks more than intentions, and more directions than backward.
If you go "no driver" then your market dies. It is going to be an ironic pain for Detroit, one that ranks up there with having your company realtime video your face for thought-crimes, when individual consumers don't buy cars that don't come with a steering wheel.
I imagine a "always try to give me the green light" patch will be created almost immediately after this system starts being used in public.
Worth following them and learning more about this.
There is still debate as to which will win out (https://www.nytimes.com/2018/09/20/business/vehicle-communic...), but GM, Toyota, and Volkswagen seem committed to DSRC. As they are the biggies, my bet is within 5 years most new cars, some new pickup trucks, and some new delivery/heavy trucks will be using DSRC.
How useful will that be?
DSRC transmits a "Basic Safety Message" 10 times per second (range 1000 meters), containing velocity vector, pitch/roll/yaw vector, vehicle weight, status ("braking", "skidding", etc) and, critically, position. Current GPS (particularly in urban canyons) may only give 10m accuracy, or worse.
But if we could get 10cm accuracy (maybe with some kind of publicly deployed differential gps?), then a whole set of advantages become apparent.
1) Your Heads-Up-Display could warn of a (DSRC-equipped) vehicle that's about to blow through an approaching stop sign and will collide with your projected position. And your auto-braking could be activated.
2) You could be warned on the freeway of a vehicle 5 cars ahead of you that has just slammed on the brakes (because that vehicle just transmitted that info).
3) Special lanes could be reserved for platooning, with cars separated by only a meter or two (0.1-0.3s) rather than 2s. Voila -- much greater lane capacity.
4) If DSRC were cheap enough ($100-200), bicyclists and motorcyclists could buy one, and make themselves much more visible to every DSRC-equipped cager.
5) And if really cheap, many pedestrians might carry them, safe in the knowledge that if they step out into the street, the auto-braking cars will pro-actively avoid them.
And _none_ of these features require any type of cameras, lidar, or other (arguably problematic) self-driving technology. Heck, they don't even require a map.
A final point about privacy -- DSRC was designed so the vehicle identifier isn't anything traceable like VIN or license. Instead, the id is randomized every few mintues (or at least according to https://pdfs.semanticscholar.org/fd03/f4c5f3128db5c08e5e92f2... -- I don't know how GM/Toyoto/VW have implemented it).
I believe this is supported/expected in the DSRC specifications. As I understand it, a well-known point, such as a post at a corner of an intersection, is located with high accuracy. A DSRC Roadside Unit can then broadcast GPS corrections to provide significantly higher accuracy to units that are able to receive and use this information.
It does require some infrastructure support, but it's something the specification group has thought of.
https://www.ncta.com/sites/default/files/2018-10/NCTA%20Fres...
IMVHO V2V coms are good ADDITION, but only addition, no more.
...which will never happen...
Although I only have a rough idea what I’m talking about, I should delve deeper to be more quantative.
Generally, I'd hope that my request to cross an intersection is confirmed in fewer than 10 minutes.
I can foresee the day when this communistic approach to traffic management will be unacceptable, and the patriotic solution is to let the market decide. Whomever is willing to pay the most gets to run all the intersections, while poorer people get to use them when there aren't any richer people in the vicinity. Because the free market always finds the optimal solution.