Fun with my car insurance company's OBD II nanny box

[230RN]

So I got one of those OBD (automotive On-Board-Diagnostic) plug-in dealies so the insurance company could see what a great driver I am.  About 25-30 years ago I learned how to actually drive, as opposed to merely "operating a car." Haven't had a ticket or an accident that was my fault since then, and only one, at that. (Knock wood.)

It monitors a couple of things, among which are braking events and night-time driving events.  They say driving between the hours of midnight and 4 AM is "bad."  OK, but since I retired, I do a lot of erranding in those hours since I'm a natural-born nightowl.   OK, so I'll have to talk to them about that.

But my second week's report showed three "evil" braking events with my comments as to what I was doing while I was out erranding as follows:

Week Jun 8 - Jun 14, 2015
6 Trips - 18.3 Miles - 1.1 Hour -
3 Braking Events
Date & Time / Duration / Description

06/14/15, 09:31 1.0 Seconds 24 - 16 Mph
^ Traffic light at Wadsworth & 38th, sudden slowdown by everybody in front of me.
8 mph/sec

06/14/15, 09:56 1.0 Seconds 37 - 30 Mph
^Downhill on 44th from radar sensor flashing at me (my bad)
  7 mph/sec


06/14/15, 09:56 3.0 Seconds 25 - 2 Mph
^Traffic light turned yellow, nobody behind me
7.66 mph/sec

The bolding shows my calculations of deceleration.

I've got a lot of questions about this whole sensing affair, like how often do they transmit to their receivers, how do they sense "speeding" on different roads, and a whole bunch of stuff I've been prowling around the net on for answers.

But a couple of things bother me right now.

I have smaller-than-standard tires on my car for unavoidable circumstances  --one standard tire got destroyed and I had to replace "it" with all four tires and all they had was smaller ones.

As a consequence, my speedometer reads about 31-32 mph when, according to the radar stations along various  streets, I'm actually going only 30.

Now I figure they can probably sense a "braking event" in a couple of ways...

(A) ...by an accelerometer built in to the nanny box.

(B) ...by brake pedal position (which I found out some OBD II systems can do) or various brake pressure readings like power brake diaphragm and brake line pressures.

(C) ...by other engine/transmission parameters such as calculations from RPM, gear ratios, manifold vacuum, etc.

(I doubt they could do it from GPS data because I don't think GPS responds that fast or that accurately.)

If it's an accelerometer in the nanny box itself, assuming proper calibration (which is sometimes questioned, according to what I picked in my amateurish search), I have no quarrel, except for the low threshold to create a braking event.  It seems too low, probably around 7mph per second.  I also wonder how they're figuring in downhill or uphill braking accelerations, where gee would have a cosine effect.

If it's by brake pedal position, the first thing I noticed when I bought the car was that the brake pedal felt low to me compared to my previous car.  However, it has not changed in the years I've been driving that car, so maybe that's the "standard" for that car.

If it's by engine/transmission parameters, then it's clearly unfair by some small amount due to the smaller tire size.  (Braking from 31-32 mph (on the speedometer) to zero will give a higher deceleration reading than from the "real" 30 mph to zero.

So, while I'm still <ahem> "researching" this whole nanny box thing, I wondered if youse* guys and gals would like to kick it around and see if youse had any hints or other comments on this stuff.

Terry, 230RN

That's Brooklynese for "you-all."

[AJ Dual]

My thought is to just make a cheap old laptop a "ODBII simulator" and rip your car's firmware and digital VIN, and just let it simulate sedate 25 mph trips to and from work, and a run to the store every week... 

[lupinus]

This is why I refuse to accept a small discount for having some random computer dictate if I am driving safely or not.

[KD5NRH]

My thought is to just make a cheap old laptop a "ODBII simulator" and rip your car's firmware and digital VIN, and just let it simulate sedate 25 mph trips to and from work, and a run to the store every week...

This is on my list of possibilities.  Also the simple powered passthrough idea that has been tossed around elsewhere, since they only capture the VIN on powerup; plug in via the passthrough, let it get the VIN, hook up a 9v battery to keep it running, take it inside and plug into a 12V wall wart.  Reconnect when it's time to go on a nice long open road drive where it will build up realistic mileage for the month and very few braking incidents.

[Firethorn]

(C) ...by other engine/transmission parameters such as calculations from RPM, gear ratios, manifold vacuum, etc.

GPS can actually give you acceleration, but I figure it's just yanking speed off the basis of the spedometer, which is run through the computer these days.  And yes, I figure that means you're getting screwed, but less so than me, my spedo is 5 mph fast @ 55, meaning it reads 60 when I'm doing 55.  

This is due to federal law that the spedo can't be slow, and I drive a truck where bigger tires are an *option*.

When I get the tires I bought used, which are bigger, I'll have to see if it drops the number of events.

[230RN]

I've seen a couple of OBD hacking sites in my so-called research, but I don't understand 90% of the jargon, so it's out of the question (besides probably being illegal).  I spent a while just trying to find out what "CAN" meant, for instance.  Sure, to those of you who deal with it very day, you know what it is offhand.

I also wonder about how much they can squeeze out of the data (although it's probably in the car's computer anyhow) if there's an actual accident.  Is it like an aircraft "black box?"

One thing I noticed is that I'm probably driving a little worse now, since I'm a little distracted by trying to think about what the nanny box is thinking about my driving. :)



I dunno, I'll play with it for a while and see what develops, and what the insurance company says about my nighttime driving habits.  If they can't change, that, it might be a deal-breaker.

AJDual said,

...just let it simulate sedate 25 mph trips to and from work, and a run to the store every week...

You forgot "and to church on Sundays."

Really a fascinating technology, all in all, and I'm still interested in comments or explanations of how they do it all.

Terry, 230RN

[MillCreek]

^^^In my risk management journals, I have read about how the manufacturer can extract a surprising amount of data from the car's black box.  I know this has been an issue in lawsuits over accidents.

[cassandra and sara's daddy]

Yea I use torque and it gets a ton


Sent from my iPad using Tapatalk

[Firethorn]

I also wonder about how much they can squeeze out of the data (although it's probably in the car's computer anyhow) if there's an actual accident.  Is it like an aircraft "black box?"

No, there's minimal recording going on, mostly some continuing optimization.  However, some newer(and generally more expensive) cars today do have actual black boxes that records something like the last 5 minutes of driving.

I dunno, I'll play with it for a while and see what develops, and what the insurance company says about my nighttime driving habits.  If they can't change, that, it might be a deal-breaker.

It probably is.  You see, those hours of the day are when the accident rate is at it's highest due to drunks & tired people on the road.

[KD5NRH]

It probably is.  You see, those hours of the day are when the accident rate is at it's highest due to drunks & tired people on the road.

And ironically, that's when probably 50% of my mileage happens, but also probably the safest driving I do; coming home from Church dances 100-150 miles away, and as soon as I'm out of DFW, I tend to be in very light traffic.

[Firethorn]

And ironically, that's when probably 50% of my mileage happens, but also probably the safest driving I do; coming home from Church dances 100-150 miles away, and as soon as I'm out of DFW, I tend to be in very light traffic.

Church dances 100-150 miles away?  How many miles do you drive a year?  You might DQ yourself from that alone.

[brimic]

Church dances 100-150 miles away?  How many miles do you drive a year?  You might DQ yourself from that alone.

Shhh! Don't tell my insurance company, but I put on 28,000 miles in the last year.

[KD5NRH]

Church dances 100-150 miles away?  How many miles do you drive a year?  You might DQ yourself from that alone.

Well, this has been an odd year, but I didn't get the device until after the 475 miles each way trip to the Rio Grande Valley.  Normally I go to 1-2 of those dances a month.  Then again, my daily commute is a bit over 6 miles each way, mom's house (once a week) is a bit under 10, and WalMart is 3, with most of the town closer than WalMart, so until I started taking the Aikido class twice a week, 40 miles away, the dances accounted for about half of my monthly mileage.

According to some hacker groups, the Progressive Snapshot uses the speedometer signal for braking, its own GPS for distance, and the VIN to verify that it's in the right car.  They say speed vs location (to check if you're speeding) isn't monitored, and odometer reading isn't a standard OBDII data point, (the cars that send it use manufacturer specific codes for it) so it isn't tracked.  Thus, the powered passthrough is a valid possibility, (keeping the VIN stored by delivering continuous power to the "always on" pin of the OBDII connector, should be hot on pin 16 and ground on pin 4, so it thinks your car is just sitting) though you'd want to store it in a fixed location within a reasonable GPS margin of error of where you remove/reinstall it, (hanging on the garage wall comes to mind) so they don't start wondering why your car travels a lot with the engine off.

Hmm...looks like the easy way to do it; make yourself a 12V supply with a female (car side) OBDII port for in the house/garage, and a temporary supply (9V battery to a male OBDII) to get from the car to the wall wart.  Plug in the passthrough in the car, plug the Snapshot into one of the ports on it.  When you're ready for an undocumented trip, plug your battery into the other port on the cable, unplug the cable, move it to your storage location and plug the cable into your 12V supply, then unplug your battery.  Power is never interrupted to the Snapshot, so it doesn't go looking for the VIN again and thinks it's just sitting in the designated car reading nothing from the computer while the GPS registers that it hasn't moved beyond normal position drift.  Reverse the process later to get it back in the car.

Presumably, if it actually requests the VIN rather than receiving it automatically at key-on or engine start or something, you could even use this process to swap it into your local little old lady's car so they only see you driving 20 miles an hour to church and bingo.  Not sure exactly how the VIN retrieval works, though, and it would be my luck to drop it in her car right before she heads off to the salt flats for a weekend of racing.

[HeroHog]

If you have ABS then you have individual wheel speed sensors on each corner and those readings can easily give you deceleration rates. Same for the spedo sensor as it is electronic these days and it's reading will give acceleration and deceleration in addition to speed just by comparing speed over time.

[KD5NRH]

If you have ABS then you have individual wheel speed sensors on each corner and those readings can easily give you deceleration rates. Same for the spedo sensor as it is electronic these days and it's reading will give acceleration and deceleration in addition to speed just by comparing speed over time.

Problem is that it's only looking for a 7mph/second drop as "hard braking" to show you're being unsafe, without considering the starting speed.  Dropping from 60 to 53 might be nailing the brakes when you see a cop, but dropping from 7 to full stop quickly is usually a pretty reasonable move; either you were already goping slow in anticipation of a need to stop quickly, or you misjudged the distance to the stop line and had to tighten up a bit.

[brimic]

There are no doubt all kinds of ways to spook the OBDII reader, but will the Ins. company cover you if you do happen to get in a serious accident, and they find that your OBDII reader is still sending data showing that you are on a leisurely drive somewhere else because it's hooked up to a laptop in your house? 
I don't think they'll take kindly to fraud.

[KD5NRH]

There are no doubt all kinds of ways to spook the OBDII reader, but will the Ins. company cover you if you do happen to get in a serious accident, and they find that your OBDII reader is still sending data showing that you are on a leisurely drive somewhere else because it's hooked up to a laptop in your house?

So keep your regular OBDII dongle plugged in, and have some trusted neighbor go unplug their device as soon as he's notified.  (Even better would be to have it plugged into an outlet you can turn off from your phone directly.)

"Well, I kept getting a check engine light and needed to run Torque on this trip to see what was wrong."

If you're a few minutes from home, I doubt they're going to check closely enough to disprove that and/or assume whatever documentation of the wreck time couldn't possibly be 5-10 minutes off.  Harder if you're a few hours' drive out, and their box says the car was sitting in the driveway at the time of the wreck.

As for spoofing it with false drive info, the best way to do that would be a GPS-equipped tablet or similar that you could carry with you, giving them data that's merely adjusted to fit the "good driver" profile, (i.e. no speeding or hard braking) but still gets you where you're going, albeit probably a little later, though a well written simulation could make up time by going over actual speed but under the limit through congested areas.  If it takes it a couple extra minutes to arrive at the accident scene, it will likely still be well within tolerance for whatever the accident report says.  Of course, this still leaves no good way to deal with "dangerous time" driving, but even simply spoofing speed data could eliminate the "hard braking" negative points.

[brimic]

Its a lot of trouble to go through to save a little money.

[Hawkmoon]

So, while I'm still <ahem> "researching" this whole nanny box thing, I wondered if youse* guys and gals would like to kick it around and see if youse had any hints or other comments on this stuff.

Of course.

Remove the device from the vehicle, set it on an earth berm in the country, and see if you can score 5 hits out of 10 rounds of 9mm from your favorite Glock.

Or just "field test" it for resistance to a 3-pound sledge hammer.

[Hawkmoon]

I also wonder about how much they can squeeze out of the data (although it's probably in the car's computer anyhow) if there's an actual accident.  Is it like an aircraft "black box?"

The newer ones are black boxes, and becoming more so every couple of years.

It was at least eight years ago, and maybe closer to ten, that my brother told me about what the "black box" in a then-current BMW could do. Guy came in with a very noisy engine. Mechanics ripped into it and found most of the valves were bent. Brother (who was the service manager) got to explain to the irate customer that the engine would not be replaced under warranty because said irate customer had been racing the car.

Irate customer absolutely swore that he always drove very conservatively and sedately -- right up until brother showed him the printout from the back box, showing that he had been exceeding the engine's redline by at least 1,000 RPM regularly.

If they could do that then, I don't want to find out what they can do now.

[griz]

This kind of thing is new to me.  Does it send you some sort of log or report?  I'm still struggling with the idea that using the brakes is a bad thing. 

[Firethorn]

This kind of thing is new to me.  Does it send you some sort of log or report?  I'm still struggling with the idea that using the brakes is a bad thing. 

I have a website I can log into that gives me a breakdown.  Right now I'm projected at a 5% discount -
A- Mileage, B- Braking, B- Time of Day, A+ speed.

Note on speed:  Don't break 80mph, that's it. 

Using the brakes isn't a bad thing, do you remember how buses stop?  That's how you do it.  Nice and gentle.  Coast to a stop, like you have all the time in the world.

Tim of day -
Lowest risk: Sat&Sun, 5 AM-11PM.
Low risk: M-F 4AM - 12PM
Moderate: 12PM - 11PM
High: 11PM - 4AM (5AM Sat&Sun)

[KD5NRH]

Using the brakes isn't a bad thing, do you remember how buses stop?  That's how you do it.  Nice and gentle.  Coast to a stop, like you have all the time in the world.

Not an option in a town with a lot of little old ladies who are convinced that as long as they never exceed 25mph, they don't have to slow down for stop signs.

[230RN]

Its a lot of trouble to go through to save a little money.

Spendthrift ! :)   Actually, I'm just investigating what it does, how, and why, like a new kid's toy.  After a while, I imagine I'll just ignore it. Installation was simply a matter of flipping open the ODB door and plugging it in.  The nannybox is specifically for your make, model, and year of the car.  There was a sealing label on the box which described my car.

griz asked,

This kind of thing is new to me.  Does it send you some sort of log or report?  I'm still struggling with the idea that using the brakes is a bad thing.

 

Yes, as I implied in my OP, there is a website to get the reported data.  Apparently, more insurance companies do it than I was aware of.  (State Farm doesn't, I asked.)  I believe the main supplier of these devices may be OCTO, www.OctoUSA.Dom. When I get enough intelligent questions together, I'm going to contact them for details, if they'll give them up.

When I first called my new insurance company about it, I was a little concerned about the "braking report," since I use my brakes (and two toots on the horn) a lot to signal pedestrians and other cars that I'm backing up, and to signal drivers behind me that there's a slowdown up ahead and the like.  They told me not to worry about that, they were only recording "hard" braking.  Mmmmkay.

But I think that maybe 7 mph/sec is a low threshold, especially if you're driving in rush hour traffic, where lane changers may suddenly cut right in front of you and the like.  (That 7mph/s is a guess from the slim report data I got right away.)  I may look at that in future reports.  The question is, who decided what was "excessive braking" and other parameters, such as what "reckless driving" and "hard turns" are and how they measure them.

So, while all this looks like "a lot of trouble," it's still very interesting how they do it all, and how they decided on the values involved in their reported events.

As I said in the OP,

I've got a lot of questions about this whole sensing affair, like how often do they transmit to their receivers, how do they sense "speeding" on different roads, and a whole bunch of stuff I've been prowling around the net on for answers.

Hence this thread.

Terry, 230RN

[griz]

It would be interesting to see that sort of information reported on one of the new self driving cars.  I suspect they would get high marks since they would program them to drive safely, but even a machine will have a "braking event" happen rather than plow in to some cell phone obsessed driver who pulls out in front of them.

That said, if you want to avoid the demerits of driving at night, perhaps you could set the car's clock from AM to PM?