I had a Volvo XC90 that “jumped” off the interstate and onto a parallel mountain road east of Knoxville. It did its best to track along those roads and somehow made its way into North Carolina. But even when I was back in Chicago, it was still stuck in NC trying to find a way off those mountain roads. Dozens of on/off cycles did nothing. I disconnected the battery overnight and that didn’t work. At the next service appointment, the dealer had to do a full firmware reset to wipe the memory and get it working again.

It amazed me that Volvo programmed an SUV to disbelieve that it could ever actually leave a road.

My guess would be:

1. Car entered ferry, loses GPS

2. Car entered dead reconning mode used for tunnels and such

3. Car left ferry, acquired GPS

Then either:

4a. Location via dead reconning vasty disagreed with GPS because the car doesn't know about the ferry's movements, triggering some kind of failsafe.

Or:

4b. There's just a plain old bug in the condition to switch back to GPS and maybe people haven't noticed because you don't get as badly desynced in a tunnel.

>the car must have a pretty good dead reckoning system

Yeah all the pieces are there: accelerometers and gyros for stability control, compass for navigation, and the wheel speed sensors give you exact distance traveled.

A practical issue is that GPS can be spoofed relatively easily. If autonomous driving became a thing, and ubiquitous, with vehicles that prioritized remote over local consistency, then a single GPS spoof could cause some interesting things to happen. This is probably a concern that does drive decisions, at least to a degree. It creates a weird scenario where you kind of need to trust GPS, but you simultaneously also can't treat it as authoritative.

OTOH creating a dumb user experience for a fairly common scenario, to try to preempt a hypothetical scenario is probably not a great idea. A fun problem to think about, though it may be completely unrelated to this issue.

> On a vaguely related note, driving 3000 kilometers through Europe in an electric car was surprisingly nice.

Having done a number of multiple-thousand km trips in Europe in an EV (not a Tesla, nobody buys those anymore) — it's amusing how non-EV muggles think this is somehow an ordeal. It's just fine! There are drawbacks: you do have to use your brain and plan ahead more than you do when burning dinosaurs. But I found that the 20-30 minute stops every 2h really improve how I feel after a day or two of driving.

Agreed about prices: there is gouging going on with some crazy margins. When charging at home, an EV is 2-4x less expensive per km than a gasoline-powered car, but when fast-charging on a road trip the cost of energy is nearly the same.

Back in the days of MapQuest there was a (usually) very good site called mapsonus.com, but it evidently had one of the ferries that crossed Boston Harbor in its map as a zero distance link.

Since it was offline, the bug was obvious although a bit frustrating - you had to put in multiple waypoints to make it forget its urge to send you on the ferry to Hull when you were trying to get to parts of the South Shore.

I suspect that my Tesla adjusts its location based on dead reckoning after losing GPS lock except that it assumes that I'm driving forwards. I.e. If I reverse out of my garage, then the map ends up in the wrong place and now I'm driving parallel to the road!

As you're mentioning a RO/RO ferry, recently, and Europe... are you sure you weren't in a GPS-denied environment, where the car (correctly) detected that GPS was jammed or spoofed and ignored it?

Civilian GPS (as in, the DoD’s Navstar) alone isn’t accurate enough to actually place you on a specific road. To compensate, auto navigation systems will snap you to the nearest road parallel to your current heading. Tesla likely didn’t consider the edge case of ferries and other off road situations in their mapping software and you hit some corner case bug.

Possibly it was using WPS keyed to the ferry's WiFi and didn't consider the possibility that the ferry itself could move out of where it was registered (ie, the port)?

One hypothesis I can offer is that your car could not, for whatever reason, use assisted gps to download the almanach which relies on mobile networks. Gps devices that download the almanac via gps itself can take up to twelve hours (in higher latitudes i think)!

My farming GPS does this every time I ship it with my vehicle to a new continent.

I always figured it has some code to get a quick location, and it assumes that location is very close to the last time it got a sat lock.

I just ignore it for the first day, then it comes good.

I wonder if it's using a compass and odometry (distance) with dead reckoning. A strange choice when GPS is available, but it would account for the map moving in the car's direction and it not changing locations when it moved without rolling (ferry).

Looks like others have reported similar issues. Entering a navigation route may help?

https://teslamotorsclub.com/tmc/threads/map-location-and-rou...

I remember cars in the early 2000's doing this. Haven't seen that behavior in years.

GPS has an altitude component so maybe it thinks either the AirTag or the tracking phone is really high up?

It’s purely a UI thing, the same (misguided) reason scrollbars disappear.

Altitude

It's unlikely that would cause a 10x increase, especially since the directions are presumably aligned to roads of non-vanishing length.

So it might be a hack to get it to take another way -- rather than invent some way of marking a road section as being closed, mark it as being 1000 miles long?

In Apple Maps? Unpossible!

Except this one is not on the windshield.

If you account for the satellite being further away because it would be above the point the distance works out okay - except that geostationary satellites go around the equator and that is a thousand miles away from that.

You wouldn't imagine ...

I have seen API calls you people wouldn’t believe.

Requests hanging off the edge of the load balancer, Ethernet tubes glittering in the dark.

Latency logs reporting some API calls that took longer than the age of the universe.

Their means and percentiles burning on the shoulders of weekly reports.

Plot points dying at the edge of y axis

... like tears in the rain.

Fixing bugs. At Apple. OK.

I think the developers didn’t account for lost AirTags many hundreds of miles away. And that long calculation was not the best.

I went back to their posts in November for this

> One of my AirTags is inexplicably 1600 miles away in Texas. It's one of my floating ones that I throw into bags when traveling as needed.

> Maybe it dropped out of my bag only to be found by someone who thought it'd be fun to confuse the owner. Or maybe it became e-waste to be dumped in another country.

The distance is more than the circumference of the Earth. If you wrapped a rope around the planet at the equator, the length would be around 25k miles.

Seems like the author of the linked post has a lost airtag located in the abyss.

Sidenote, I love spotting random binary numbers out on the wild

1.2 times the circumference of the Earth. Is space compressed between SF and Guatemala City somehow?

Settings - privacy - location services - system services - compass calibration. Make sure it’s enabled. In a bout of paranoia I disabled location for everything I didn’t recognize as necessary and then it took me many months to find out why compass cone behaves funny in mapping apps.

On my past iPhones, the compass always got screwed up over time somehow. Sometimes recalibrating fixed it, but only for a minute.

That's a phone compass problem no? Perhaps the specific way you're holding the phone is confusing for the algorithm?

The flight distance is 2,520 miles, Google puts road distance at 3,179 mi = 1.26x, 29,905 mi is 11.87x. Road miles are not even close to accounting for it, it’s just a bug.

> If you put all of your blood vessels end to end they would go to the moon and back.

The moon orbits too quickly to make this practical

Being a little bit clever can lead you to make some pretty bad mistakes.

Yes, the distance according to roads can be different from the distance as the crow flies. No, it cannot realistically be 10x the distance when the crow's distance is 2500 miles.

You would have to take an implausible number of wrong turns to have a chance of making it anywhere near that long a drive. For an example of a shorter but still entirely unreasonable path: Head straight north all the way to the pole, continue south to the pole, continue north to the latitude of the destination, then head due east until arrival.

I question your use of the word “easily”.