Charging locations grid based and related problems.

[michael]

As WattsUp reported some time ago and several other of us have noted, the location based charging system seems to have been grid based. This resulted in the car being treated as if at some "ghost" locations adjacent to the actual charging location.

I saw numerous examples of this and have discussed it at length with the Ford tech people. They tell me they have looked into this issue and believe they have fixed or at least improved the situation.

It appears to me they have been at least partly successful I cleared out my list of charging locations and so far only one "ghost" location has appeared.

Could anyone else who is interested in this please look into it? See if you think it's improved or not? I told the Ford people I would check with the other forum members and let them if whether they have fixed this issue.

Thanks

 

[WattsUp]

Interesting... I will check things out and report back here.

For those interested, here is my original posting from December 2013:

MFM charge locations seem grid-aligned... causing problems

 

[brogult]

I don't use value charging so I don't worry about my charging location, but I did notice something the other day that might possibly be related.

I was driving down a street and had the map display up. It showed me driving what I guess to be 100 feet off to the side of the road--terrorizing pedestrians and knocking over trash cans, I suppose. Then I made a right turn and the error persisted--it showed me ahead of where I was, not by a huge amount, but well outside its normal precision. The error was constant direction-wise--it showed me about 100 feet south of my actual position. This continued until I parked the car and went into a store, then when I came back out and drove away, it was normal.

GPS can make errors due to signal interference and reflections or multipath, but the odds of that error persisting through multiple readings on a moving platform are impossibly low, especially when you change directions. I'm assuming this is a software problem somewhere in the navigation system and whatever it is, it might be your ghost as well.

 

[WattsUp]

Interesting... I will check things out and report back here.

For a second there I had some hope, but the initial results are not promising.

I erased my home and work charge locations and then charged (at home). The newly-recorded charge location is still about 100 feet off of the true location of the car (in my garage) and appears to be in exactly the same old "grid aligned" location as before.

I will collect a few more data points (when I charge at work tomorrow) but my impression is that the Ford engineers haven't changed anything fundamental regarding the grid-alignment behavior (which, IMO, is the problem).

 

[michael]

Thanks. That will be helpful

I erased everything, only one ghost location in about three months but I haven't been using as many different actual locations as previously.

 

[michael]

I received another callback from Ford. They are still looking into this. They asked me to clear my list of locations and see if any ghosts re-appear.

Anyone else interested in the problem, could you please do likewise? I'm no longer charging at a wide variety of locations so my report to them won't be conclusive.

 

[WattsUp]

After a couple/few weeks charging around the various lots at work, all same grid-aligned locations have appeared again. Nothing seems different.

 

[michael]

OK, thanks. I'll tell them when they check back.

Perhaps you could start your own issue with them? What I did was enable them to look at the stored charging locations in my file, let them see the "ghost" locations for themselves.

 

[michael]

I had another discussion today with the Ford rep about this issue. We discussed the idea that the charging location system should be able to recognize that if there's a spot you use frequently and the system thinks it's a few hundred feet away, it should be smart enough to realize that this is a location estimation problem, not a separate charging location.

Separately, he told me something so surprising I am skeptical. He said the charging location system is cellular based, not GPS based. And as ATT removes 2G capability, the vehicle location and status reporting system is becoming less reliable.

I find it surprising that GPS isn't used for location sensing. If that's true, no surprise all these ghost locations pop up. I wonder if what he told me is correct?

 

[hybridbear]

I had another discussion today with the Ford rep about this issue. We discussed the idea that the charging location system should be able to recognize that if there's a spot you use frequently and the system thinks it's a few hundred feet away, it should be smart enough to realize that this is a location estimation problem, not a separate charging location.

Separately, he told me something so surprising I am skeptical. He said the charging location system is cellular based, not GPS based. And as ATT removes 2G capability, the vehicle location and status reporting system is becoming less reliable.

I find it surprising that GPS isn't used for location sensing. If that's true, no surprise all these ghost locations pop up. I wonder if what he told me is correct?

I don't think it's that surprising that the AT&T cellular modem is used for the MFM location and not the car's navigation. We've found that there are certain locations where there is LTE AT&T service on our iPhones, but there must be no 2G AT&T service because the car still says "out & about" even though it is off and parked. The MFM status does not update until you start the car and drive out of that neighborhood, then it works again.

 

[michael]

The reason I found it surprising is that GPS is so much more accurate and works even without cell service. Yes, to report the location needs cellular data, but not to find where the car is stopped. Why would they do that?

 

[WattsUp]

Separately, he told me something so surprising I am skeptical. He said the charging location system is cellular based, not GPS based. And as ATT removes 2G capability, the vehicle location and status reporting system is becoming less reliable.

I think that's crap. (Directed at the Ford engineer, not you.) That means you couldn't use navigation when you have no cellular signal, yet you can. The car almost certainly locates itself via GPS. Cellular-based location would be fairly inaccurate.

Also, before the big BEV/PHEV site migration/combination a couple years ago, the charging locations used to be much more accurate. (The icons for charging stations used to appear nearly "right on top" of the icon for your car on the MFM map.) This alone demonstrates that what that engineer said is coming right out his butt.

After the migration, something changed (I believe) about the server-side software and it started "grid-aligning" all the charging locations. So, although the system does know an accurate position for your car, for whatever reason, the charge locations recorded by the system are aligned to the nearest grid crossing (about 200 feet by 250 feet). This is why, if you happen to park near the middle of two grid crossings (either vertically or horizontally), due to slight GPS variation, the system might choose one or the other.

 

[michael]

I think you are right in all respects. Just reporting what he said.

 

[hybridbear]

That means you couldn't use navigation when you have no cellular signal, yet you can. The car almost certainly locates itself via GPS. Cellular-based location would be fairly inaccurate.

I don't think that would necessarily be true. I think he might be indicating that there are two separate location systems in the car. One for the navigation system which is installed on all Energi/Electric vehicles, even those without navigation (the Fusion Energi does not have navigation standard as in the C-Max Energi & Focus Electric but it still has the GPS system for EV+ mode). But the TCU could use the AT&T cellular radio to determine the location which appears on MFM.

Anyone have an idea of how we could test this to know if there really are two systems? Maybe checking something when the car thinks its location is goofy on MFM? In the underground garage at our apartment the MFM location is usually accurate, but once in awhile the location goes screwy. If we flip to the navigation screen in the Focus before leaving the garage it usually shows our location accurately, but sometimes it is also incorrect. I'll have to pay attention to see if these errors happen at the same time which would indicate that there is only one location system.

 

[jmueller065]

One for the navigation system which is installed on all Energi/Electric vehicles,...

Go bigger: on all MyFordTouch vehicles--which today is all Fords. The MyFordTouch module is the same spec'd module in all Fords (there are different vendors for different vehicle lines but the module spec is differed--for instance the MFT display in my wife's Escape is from a different vendor than the one in my FFE).

Anyone have an idea of how we could test this to know if there really are two systems? Maybe checking something when the car thinks its location is goofy on MFM? In the underground garage at our apartment the MFM location is usually accurate, but once in awhile the location goes screwy. If we flip to the navigation screen in the Focus before leaving the garage it usually shows our location accurately, but sometimes it is also incorrect. I'll have to pay attention to see if these errors happen at the same time which would indicate that there is only one location system.

Sure: GPS cannot be received inside garages, etc. whereas cellphone signal can.

 

[WattsUp]

But the TCU could use the AT&T cellular radio to determine the location which appears on MFM.

Yeah, but the best you could do with one cell tower is know that the car is somewhere within signal reach of that tower. That would only give a very approximate location of the car.

You would need at least three cell towers and highly calibrated clocks in all of them to triangulate the car's position with any degree of accuracy. That is basically how GPS works. But, GPS is only as accurate as it is because it employs perfectly synchronized atomic clocks (on-board the satellites) and even takes calibrations for the minute effects of general relativity into account.

I find it hard to believe that the FFE is using cell towers for the car location for any purpose. It would be too inaccurate and unreliable.

 

[WattsUp]

Sure: GPS cannot be received inside garages, etc. whereas cellphone signal can.

Again, cell phone signals are not a magical substitute for GPS. They can provide an approximate location only (and only when cell signals are present). In the general case, the accuracy is nothing like that of GPS.

In any case, most GPS systems, when the GPS signal is lost, resort to inertial sensors to judge the car's relative travel from the last known point for short distances. This is meant to cover cases where the car temporarily goes into a tunnel, garage, or urban canyon. It isn't perfect, and sometimes estimates the wrong location (or just gives up) if the signal has been lost for too long.

 

[jmueller065]

Again, cell phone signals are not a magical substitute for GPS. They can provide an approximate location only (and only when cell signals are present). In the general case, the accuracy is nothing like that of GPS.

In any case, most GPS systems, when the GPS signal is lost, resort to inertial sensors to judge the car's relative travel from the last known point for short distances. This is meant to cover cases where the car temporarily goes into a tunnel, garage, or urban canyon. It isn't perfect, and sometimes estimates the wrong location (or just gives up) if the signal has been lost for too long.

I wasn't implying that cell phone signals are magical replacements for GPS signals. I was merely suggesting a testing methodology: GPS signals are easily blocked and hence a way to test. Fix one variable and try again; basic debugging--I do it all day.

 

[michael]

While "network based" location isn't as good as GPS, it's surprisingly good. I looked into it as a result of this discussion.


I found a paper

http://www.intrans.iastate.edu/publications/_documents/midcon-presentations/2009/WangCellular.pdf


Which said in part:


According to the requirement by the FCC (2001), location accuracy and reliability should be 100 meters for 67% of calls and 300 meters for 95% of calls for network-based solutions; 50 meters for 67% of calls and 150 meters for 95% of calls for handset-based (GPS-enabled) solutions. Depending on the technology, calculation methodology, and signal path, the accuracy of cell phone location estimation are varied.


For a network-based system, two major methods (as shown in Figure 1) are used to calculate the location of the cell phone. The first one is a triangulation method. In ideal conditions, the cell phone location can be calculated exactly using the triangulation method w
ith computed distances from three nearby stations.However, in reality, the computed distances are dependent on the reflections, diffraction, and multipath occurrences of the phone signal. The triangulation method result is an area instead of a point. The accuracy of triangulation method is about 50–200 meters (Openwave, Inc. 2002)


So cell phone triangulation give location to 50-200 meters, typically....this is consistent with the type of mis-location we are dealing with here. So the guy at Ford may have been telling me correctly.


That paper goes on to say:


The other method is an angle of arrival method. In this method, special antenna arrays are installed at the base stations to calculate the direction the signal.
Thus, two stations are enough to calculate in what direction the cell phone signal is coming from. Considering the effects of multi-propagation, this method has an accuracy of about 50–300 meters (Openwave, Inc. 2002).

For handset-based systems using the GPS satellite system to calculate the position of the cell phone, the accuracy is between 5 and 30 meters (Openwave, Inc. 2002). The accuracy is affected by factors such as the ionosphere, troposphere, noise, clock drift, ephemeris data, multipath, etc.

When you think about it, it does make some sense...a lot of cars don't have a Nav system, and therefore may not have a GPS receiver.

 

[WattsUp]

So cell phone triangulation give location to 50-200 meters, typically....this is consistent with the type of mis-location we are dealing with here. So the guy at Ford may have been telling me correctly.

Possibly, but I think there is an important difference to point out...

The "mis-alignment" of charging locations manifests as an exact and repeatable grid alignment, wherein the charging locations recorded by MFM always vary from the actual charging location as the result of an apparent alignment to a fixed grid. This suggests to me that the mislocations are the result of deliberate quantization (rounding) of the otherwise accurate GPS coordinates, as opposed to "random error" that could be attributable to cell signal triangulation (though I still highly doubt triangulation is even being used).

In my case, when I've removed previously-recorded charging locations and waited for them to re-appear (after charging there again), they always re-appear in exactly the same grid-aligned locations. As a test, I also observed that the alignment behavior persists across an account reset. Such behavior does not appear to me to be due to randomness -- it appears to be deterministic, and therefore is very likely to be intentional and algorithmic, IMO.

I believe the MFM server-side software is specifically doing this. It is not just happening due to error.