Off topic: NHTSA says V2V tech will save a min of 439K crashes per year, 13%

[DanCar]

https://www.nhtsa.gov/sites/nhtsa.dot.gov/files/documents/v2v_pria_12-12-16_clean.pdf
From page 10:
Undiscounted, at full V2V adoption, the proposed rule would

• Prevent 439,000 to 615,000 crashes (equivalent to 13 to 18 percent of multiple light vehicle crashes)
• Save 987 to 1,366 lives

I'm thinking this will make driverless cars easier to implement, especially if it were extended above V2V to something like pedestrian (phone) to vehicle and traffic lights to vehicle.

 

[S4WRXTTCS]

How is this off topic?

V2V tech is very integral to self-driving cars, and the article also says that.

 

[DanCar]

Yeah I regretted saying it was off topic, after adding the comment how it relates to self driving cars, but the forum software didn't let me change the title at that point.

 

[eli_]

Anytime I hear V2V described it sounds like some sort of throwback to the 1990s, like before cellular networks and the Internet became pervasive. I don't understand why they want to invent this completely new communications infrastructure and waste a ton of spectrum on it. Just put a cell modem in the car and throw up a message queue on a server, problem solved. It'll probably be standard equipment before long just for infotainment stuff.

 

[S4WRXTTCS]

Anytime I hear V2V described it sounds like some sort of throwback to the 1990s, like before cellular networks and the Internet became pervasive. I don't understand why they want to invent this completely new communications infrastructure and waste a ton of spectrum on it. Just put a cell modem in the car and throw up a message queue on a server, problem solved. It'll probably be standard equipment before long just for infotainment stuff.

Cell service doesn't have 100% coverage.

Plus V2V is limited in terms of impact range. If someone wanted to mess with the system they could only mess with a small local area versus messing with the entire country. Also, V2V is commonly used in other countries and all it really does is harmonize us with what other countries are doing.

I like it from the standpoint of toll collection in that it gives toll operators a single standard to toll versus many standards.

I also like it from the standpoint of adding intelligence to stop lights. Everything from letting the car known how long the green will be (to optimize speed) to better traffic management.

 

[dgatwood]

Anytime I hear V2V described it sounds like some sort of throwback to the 1990s, like before cellular networks and the Internet became pervasive. I don't understand why they want to invent this completely new communications infrastructure and waste a ton of spectrum on it.

Because without an inherently untrusted and untrustworthy system for allowing one vehicle to scare another vehicle into doing something dangerous, the rise of autonomous vehicles will make it almost impossible to commit political assassinations by causing car accidents. Only if a communication system makes it possible to hack another vehicle or otherwise convince it that there is some obstacle that it absolutely must evade quickly can you feasibly trick two vehicles into crashing into each other in a way that would cause one of them to flip or drive off an embankment.

Whoops. Did I say that out loud? What are those black vans doing outside?

But in all seriousness, every time I hear "V2V", I immediately think "clueless politician who doesn't understand computers or computer security," and then the insanity suddenly makes a lot more sense. To someone who doesn't understand computers, the idea of one car being able to communicate risks to other cars in the fleet sounds like it should be a good idea.

The problem is that V2V can't feasibly cause a computer to react sooner than when it sees the brake lights of the car in front of it, and unless vehicles are tailgating (translated "being driven by humans"), it shouldn't matter if the vehicle two cars ahead has to brake, because there should be enough time for the message to propagate via a chain of brake lights. Therefore, there is no feasible scenario where V2V could prevent even one accident per year, much less 439,000 accidents per year.

Okay, maybe, just maybe you might occasionally be able to warn about black ice in some spot, and if vehicles aren't programmed to slow down in curves when the temperature is at or below freezing, then maybe, maybe you might occasionally prevent an accident or two, but only in a very narrow time window, and only if you have a near continuous flow of cars behind you, all within range of the V2V system. Of course, if you have that, chances are you also won't have black ice accumulation, because it tends to mostly form on roads that are not heavily traveled. After all, within minutes, that car will reach a non-cellular-dead zone and could communicate that information in a much simpler fashion using existing hardware, where any other car in the fleet could pick it up and remember it for when it enters that area in the future. So maybe single-digit accidents? Maybe even double-digit?

So basically, that fairly significant extra amount of engineering and the significant extra hardware costs (not to mention extra points of failure) would add almost no additional safety, in practice. It would, however, create an amazing opportunity for hacking and malicious mischief. It's a security hole so big that you can drive a truck through it — literally, if you try hard enough.

But if you assume that the person suggesting it still thinks that the Internet goes down whenever his/her cable line gets corroded, still uses the phrase "world-wide web", calls email messages "internets", and just got his/her computer de-virused for the fourth time this month, you can at least understand why such a person would want to mandate such a disaster waiting to happen.

I like it from the standpoint of toll collection in that it gives toll operators a single standard to toll versus many standards.

I also like it from the standpoint of adding intelligence to stop lights. Everything from letting the car known how long the green will be (to optimize speed) to better traffic management.

Both of those are technically I2V/V2I (infrastructure to vehicle/vehicle to infrastructure) communication. And as long as it is being used exclusively in an advisory fashion (to optimize speed), with primary detection done optically, the risk introduced is probably pretty small. Of course, once all vehicles are autonomous, traffic lights likely won't be needed anyway, so I remain unconvinced.

And paying tolls with toll tags is cheap, doesn't require any changes to vehicle hardware, and doesn't introduce any security risks. I can't see why anyone would willingly make that be part of the vehicle. Also, the tracking concerns with doing so are pretty significant.

 

[S4WRXTTCS]

And paying tolls with toll tags is cheap, doesn't require any changes to vehicle hardware, and doesn't introduce any security risks. I can't see why anyone would willingly make that be part of the vehicle. Also, the tracking concerns with doing so are pretty significant.

In WA state we have these things called Flex Passes that allow a person to switch between HOV mode (when you have multiple people in your vehicle) to non-HOV (when you don't have enough people to qualify)

That type of system simply doesn't work with basic toll tags. Plus toll tags along with the Flex passes have issues when placed behind windshields with metal in them.

So I'd much prefer to have that kind of system inside the car where it's fully integrated, and universal across wherever I drive.

As to tracking concerns I fail to see how it's going to add anything as your license plate gets constantly read anyways. Plus in the shift to EV's we're going to have to come up with a new way to take road use. The old way of taxing gas no longer works, and the EV fee doesn't take into consideration the massive difference in amount of miles driven by one person versus another person.

If it was me I'd make every car on the road have a V2X type system that was at the very least passive.

Oh, and also I don't know if you're aware of this or not, but Apple is likely going to release there bluetooth tagging system fairly soon. It's similar to the Tile. This might not seem like that big of a deal, but it is because of the massive install base of iPhones. So any iPhone (with bluetooth enabled) can report the approximate position of a bluetooth tag being searched for. Which basically means if anyone wants to track anyone all they have to do is slide one of these onto the vehicle/bike/etc of the person they want to stalk.

The Tile wasn't much of a tracking system because it relied on the Tile install base which isn't anywhere close to that of the iPhone install base. That's why it never worked all that well to try to track stolen items.

 

[dgatwood]

In WA state we have these things called Flex Passes that allow a person to switch between HOV mode (when you have multiple people in your vehicle) to non-HOV (when you don't have enough people to qualify)

That type of system simply doesn't work with basic toll tags. Plus toll tags along with the Flex passes have issues when placed behind windshields with metal in them.

So I'd much prefer to have that kind of system inside the car where it's fully integrated, and universal across wherever I drive.

Except that every area has different rules with different requirements. For example, do you set the toll tag to multiple passengers when you have an EV? Some areas, yes, some no. And how does the car know whether the buckled seat belt in the back with minimal weight is a child carrier or a 35-pack of bottled water?

As to tracking concerns I fail to see how it's going to add anything as your license plate gets constantly read anyways.

Cost/complexity. It takes negligible effort to power up a device that watches for nearby vehicles that are continuously transmitting an identifier, and negligible effort to store that data. Doing it with cameras and license plate readers requires camera hardware pointing at every lane, a decent amount of CPU power (and thus much higher electrical power requirements), and a high risk of missing cars that are too close together. Governments can do it easily, but it is nontrivial for anybody else to do it without getting noticed.

And worse, because it is tied to the car, there would likely be no way to change your identifier.

Oh, and also I don't know if you're aware of this or not, but Apple is likely going to release there bluetooth tagging system fairly soon. It's similar to the Tile. This might not seem like that big of a deal, but it is because of the massive install base of iPhones. So any iPhone (with bluetooth enabled) can report the approximate position of a bluetooth tag being searched for. Which basically means if anyone wants to track anyone all they have to do is slide one of these onto the vehicle/bike/etc of the person they want to stalk.

With very limited range, requiring you to target only a specific individual, and easily defeated if discovered.

The Tile wasn't much of a tracking system because it relied on the Tile install base which isn't anywhere close to that of the iPhone install base. That's why it never worked all that well to try to track stolen items.

No, the reasons it never worked were:

• Limited range. The odds of a phone that knows that a tag is missing being within a hundred-foot (or even 200-foot) range are remarkably low.
• Limited life. A device that must be replaced every year has a decent chance of being dead when you need it.
• Limited ability to distribute lists of missing items.
• Obvious and easily removed.

That last one is not a solvable problem. Your phone isn't going to use the power to contact a server somewhere and report that it detected a tile just because it did. That's a huge waste of power, and if they become popular enough, it will add up. Nor is the opposite approach — distributing a list of every missing device to every phone in an entire country — plausible, for the same reason.

The only truly viable tracking approach for stolen gear involves the device itself checking in with some server via cellular communication once a day, and if the device has been reported stolen, obtaining its best-guess location using GPS and server-based Wi-Fi signal triangulation and uploading that data as well. And even that works only if the thieves don't notice it and dispose of the tracker.

A better solution to equipment theft would be a law requiring all equipment costing more than $50 to provide a passcode lock that can be reset only by the original owner or the manufacturer. It has worked reasonably well for cell phones.

 

[S4WRXTTCS]

Limited range. The odds of a phone that knows that a tag is missing being within a hundred-foot (or even 200-foot) range are remarkably low.

The odds of the phone knowing a tag is missing is a function of it's range, and it's install base. The tile app itself takes care of knowing if a tag is being looked for.

Tile really is a victim of Apple, and Google not incorporating support for them within the app itself.

In public how often are you within cell range, and not within 200ft of an Apple phone? Apple recently added bluetooth tracking to find missing iPads, and MacBooks. I've been meaning to test this to verify that it works because their tag is going to use the same mechanism, but I haven't gotten around to it.

I can't see a GPS based tracker being anymore effective than this. I say this as someone who has been playing around with a custom bike tracker (based on particle.io boards), and it's convinced me that there is no need for me to continue on with the project. That there won't be any demand for it.

Now limited life is a huge issue with bluetooth trackers, but that's where product integration comes in. Tile has had very limited success with this. It remains to be seen if Apple will be more successful. This is critical because product integration allows the product itself to power the tracker. This also addresses the easy to see, and easy to remove. I didn't have much issue with this with a tile because it's pretty small. I hid one behind the light, and one under the seat of my bike. I'm sure I could have made it even more hidden if I wanted to.

 

[dgatwood]

The odds of the phone knowing a tag is missing is a function of it's range, and it's install base. The tile app itself takes care of knowing if a tag is being looked for.

When I’m looking for my TV remote in California, it does no good for phones in Alaska to know that it is missing. The cost for every cell phone to know about every missing tracker scales linearly with the number of trackers sold, which means it rapidly becomes infeasible for every phone that might be near the tracker to know that the tracker is missing unless it belongs to the same owner, or at least is fairly local.

And for stolen stuff, the odds of it being in another city approaches 100% within 24 hours—another state within three days. So the fanout is a big problem, particularly if you’re talking about potentially tens of thousands of missing devices. Doing the lookup alone would bring the app to its knees.

GPS works because the tracker only needs to poll the central server to report its location, and that server knows if it is stolen and can change the polling interval to get data more often. That isn’t likely to be possible if the tracker has no connectivity, because either it will eat a colossal amount of battery power on the phone to report in quickly (in which case most users will turn off the feature) or it will be delayed so much that the phone won’t be in range to notify the tracker that it is stolen, so future devices won’t have any way to know that they need to immediately report that device when seen.

Also, when was the last time someone with stolen goods left them somewhere that was within double-digit feet of other people’s cell phones? (Remember that range falls off quickly through walls.) Lost stuff, maybe. Stolen? No.

 

[S4WRXTTCS]

When I’m looking for my TV remote in California, it does no good for phones in Alaska to know that it is missing. The cost for every cell phone to know about every missing tracker scales linearly with the number of trackers sold, which means it rapidly becomes infeasible for every phone that might be near the tracker to know that the tracker is missing unless it belongs to the same owner, or at least is fairly local.

And for stolen stuff, the odds of it being in another city approaches 100% within 24 hours—another state within three days. So the fanout is a big problem, particularly if you’re talking about potentially tens of thousands of missing devices. Doing the lookup alone would bring the app to its knees.

GPS works because the tracker only needs to poll the central server to report its location, and that server knows if it is stolen and can change the polling interval to get data more often. That isn’t likely to be possible if the tracker has no connectivity, because either it will eat a colossal amount of battery power on the phone to report in quickly (in which case most users will turn off the feature) or it will be delayed so much that the phone won’t be in range to notify the tracker that it is stolen, so future devices won’t have any way to know that they need to immediately report that device when seen.

Also, when was the last time someone with stolen goods left them somewhere that was within double-digit feet of other people’s cell phones? (Remember that range falls off quickly through walls.) Lost stuff, maybe. Stolen? No.

That isn't how Apples implementation works.

Here is some info on how Apple implemented the "find my" using bluetooth.
Apple's 'Find My' Feature Uses Some Very Clever Cryptography

As I understand it all the heavy lifting of matching tags with tags being looked for happens on Apples servers. That might seem like a lot of data being uploaded, but I doubt the average iPhone will encounter that many tags on an average day. Maybe a few hundred for someone out and about. So there isn't that much data or that much CPU processing power.

Here is the story of it working at least for one person. This is assuming it wasn't connected via Wifi.
New iOS 13 'Find My' app helps Redditor recover iPad lost at airport

As to GPS tracking with cellular connectivity I loved the idea in theory. The problem was I often had issues getting good connectivity on both devices. The GPS wants to be out in the open, and struggled working inside a building. It was on a bike so I didn't worry too much about it. The other issue is cell connectivity as my house was on the edge. My phone usually kept one or two bars (at most), but the tracker not so much. The other thing that was frustrating is waiting for service plans for internet of things devices which is only finally now happening.

I haven't given up completely on it, but I think Apple ruined my plans to build a bike tracker. So instead I've decided to work on a different type of bike lock.

 

[dgatwood]

That isn't how Apples implementation works.

Here is some info on how Apple implemented the "find my" using bluetooth.
Apple's 'Find My' Feature Uses Some Very Clever Cryptography

Which just adds another layer of problems.

As I understand it all the heavy lifting of matching tags with tags being looked for happens on Apples servers. That might seem like a lot of data being uploaded, but I doubt the average iPhone will encounter that many tags on an average day. Maybe a few hundred for someone out and about. So there isn't that much data or that much CPU processing power.

If you're only encountering a few hundred of them, then that means nobody is using them, and the product has failed. Also, remember that every iOS device is a beacon. If you only encounter a few hundred of those, then the entire Apple brand has failed.

Here is the story of it working at least for one person. This is assuming it wasn't connected via Wifi.
New iOS 13 'Find My' app helps Redditor recover iPad lost at airport

Sure it works, but it works by batching up all the beacons that you detected and sending them out maybe once a day. In one of the world's busiest airports, it got two hits in a day. Mind you, iOS 13 probably had low-double-digit adoption at that point, so in a really busy location, you might even get a ping every hour these days.

That's fine for finding things that are lost. It has no value for things that are stolen, where they're likely to be in an area that has only a few people, the odds of getting a ping before it is long gone are approximately zero.

As to GPS tracking with cellular connectivity I loved the idea in theory. The problem was I often had issues getting good connectivity on both devices. The GPS wants to be out in the open, and struggled working inside a building. It was on a bike so I didn't worry too much about it. The other issue is cell connectivity as my house was on the edge. My phone usually kept one or two bars (at most), but the tracker not so much. The other thing that was frustrating is waiting for service plans for internet of things devices which is only finally now happening.

That's why any GPS tracker should periodically check its location and store it, rather than doing it only on demand. It won't tell you where in a building the tracker is, but it will still give you a pretty good idea which building it is in.

 

[S4WRXTTCS]

If you're only encountering a few hundred of them, then that means nobody is using them, and the product has failed. Also, remember that every iOS device is a beacon. If you only encounter a few hundred of those, then the entire Apple brand has failed.

Wait, what? Why would I personally encounter more than a few hundred a day? I'm not within bluetooth range of THAT many people per day.

Sure it works, but it works by batching up all the beacons that you detected and sending them out maybe once a day.

Where are you get the batching up data? I don't think that's published anywhere. It will be interesting to know if it's batched, and what the reporting rate is.

where they're likely to be in an area that has only a few people, the odds of getting a ping before it is long gone are approximately zero.

I think your overestimating criminals. Most criminals are likely junkies trying to get a fix. Pretty much everything I've ever had stolen has been this way where they quick pawn/sell a device they stole.

The biggest issue likely won't be finding the device, but getting help from the police in getting the device before it gets scrapped for parts.

That's why any GPS tracker should periodically check its location and store it, rather than doing it only on demand. It won't tell you where in a building the tracker is, but it will still give you a pretty good idea which building it is in.

I set mine to automatically wake up every X number of hours to send the position data. One of the functions I could send to it was to increase the reporting rate in the event I needed to find it. The other thing that's cool about the particle.io stuff is you can program it through the cloud so you can completely change the programming if you need to.

In any case I think we're way off topic now. Luckily the OP marked the entire post as off-topic.

 

[dgatwood]

Wait, what? Why would I personally encounter more than a few hundred a day? I'm not within bluetooth range of THAT many people per day.

Within BTLE range (a couple of hundred feet)? An hour in a shopping mall will get you more pings than that.

Where are you get the batching up data? I don't think that's published anywhere. It will be interesting to know if it's batched, and what the reporting rate is.

I can't imagine that it wouldn't be. Turning on the cellular radio is relatively expensive, and if you did it every time you walked past any person with an iPhone, it would be like talking on the phone for the entire duration of a typical walk down a busy city street.

I think your overestimating criminals. Most criminals are likely junkies trying to get a fix. Pretty much everything I've ever had stolen has been this way where they quick pawn/sell a device they stole.

At least here in the Bay Area, most theft from automobiles (and Teslas are being hit harder than average, so this is perversely on-topic) is actually organized crime. There's a scout that comes by on a bicycle and figures out which cars to hit, but never does any actual theft, and then somebody else comes by in a vehicle, breaks the window with one of those glass breaking tools, pulls out stuff, throws it in the vehicle, and drives away. Police have found warehouses full of stolen goods in various cities, most of which was stolen from other cities. This isn't petty crime anymore; it's big business.

 

[S4WRXTTCS]

An hour in a shopping mall will get you more pings than that.

Shopping mall? What decade do you think I'm in? The 80's??

I can't imagine that it wouldn't be. Turning on the cellular radio is relatively expensive, and if you did it every time you walked past any person with an iPhone, it would be like talking on the phone for the entire duration of a typical walk down a busy city street.

I agree that there needs to be some level of batching, but that's not what I asking. I was asking if you had any information as to how much is batched, and how often it is reporting location data.

Police have found warehouses full of stolen goods in various cities, most of which was stolen from other cities. This isn't petty crime anymore; it's big business.

Being collected like that is probably true of most theft these days. Like a friend recently had her bike stolen, and the police ended up finding it near a homeless encampment with lots of other bikes. Basically they'd bring the bikes together, and they'd mix parts. She did get her bike back, but it had a different handlebar along with a different seat. But, see this works to their disadvantage because all it takes is one or two stolen phones that haven't been turned off to reveal the stash to police.

What I think is interesting about the Apple airtag is it's the first time that I know of that will see a mix of device to device (like V2V) communication along with device to cloud on a massive scale.

 

[dgatwood]

I agree that there needs to be some level of batching, but that's not what I asking. I was asking if you had any information as to how much is batched, and how often it is reporting location data.

No, no idea.

 

[electronblue]

Anytime I hear V2V described it sounds like some sort of throwback to the 1990s, like before cellular networks and the Internet became pervasive. I don't understand why they want to invent this completely new communications infrastructure and waste a ton of spectrum on it. Just put a cell modem in the car and throw up a message queue on a server, problem solved. It'll probably be standard equipment before long just for infotainment stuff.

Sounds like a privacy nightmare. Not saying it won’t happen, just that privacy is one upside of local V2V.

 

[electronblue]

Because without an inherently untrusted and untrustworthy system for allowing one vehicle to scare another vehicle into doing something dangerous, the rise of autonomous vehicles will make it almost impossible to commit political assassinations by causing car accidents.

Why would V2V have to increase risks? Obviously the cars could and should still use their other senses to make the best overall judgement of a given situation.

If there is a V2V message of a hidden car approaching from the side, slow down extra. If not, you still slow down when you see it or when traffic rules or driving policy require, as normal. All you lost is the extra slowing down, but still you would have normal slowing down.

 

[dgatwood]

Why would V2V have to increase risks? Obviously the cars could and should still use their other senses to make the best overall judgement of a given situation.

If there is a V2V message of a hidden car approaching from the side, slow down extra. If not, you still slow down when you see it or when traffic rules or driving policy require, as normal. All you lost is the extra slowing down, but still you would have normal slowing down.

Any time you introduce a new communication protocol and new hardware, you add security risk. And when it is designed by a committee of automakers instead of tech companies, multiply that risk by 1000.

 

[electronblue]

Any time you introduce a new communication protocol and new hardware, you add security risk. And when it is designed by a committee of automakers instead of tech companies, multiply that risk by 1000.

Sure, the communication protocol itself certainly is open to risks, but the car would still have to make a holistic determination regarding what to with that information and could consider the data suspect in the sense that it would only take non-dangerous, defensive action due to it, which would result in increase of safety, even if it did increase the occasional false positives.