Glitches in Garmin V8 - open letter to Garmin

Wapping

Well-known member
Moderator
Joined
Oct 21, 2005
Messages
93,597
Reaction score
19,914
Location
Wapping, London
On the whole I am very happy with my BMW Navigator II and associated Garmin products, V8 and all the software updates correctly installed. The level of mapping and detail in Europe is now very good.

Leaving aside the oddity over the device’s / software inability to plot certain long distance routes, where the BMW device ran out of memory (see earlier post) I have found some real problems with V8 that did not seem to be so common on V6 or V7. These were not one-off problems and were not limited to my device alone. All of us using Garmin based navigation on the same routes, at the same and, independently, on other occasions on different routes, had the same problems. Here is my go at explaining what the problems are:

We are all familiar with the ‘spoken instructions’ not matching the screen / or what the driver actually sees on the road. These are usually limited to the software interpreting a hairpin (or similar sharp or long bend in the road map) as a junction (or maybe a fork) in the road. Hence, the spoken instruction to sometimes, ‘Turn left’ or ‘Bare left’ or 'Keep left’ on what is otherwise a simple route. We are also familiar with errors in satelite reception sometimes 'shifting' the vehicle's position off the road into a field. These I can live with.

What V8 seems to be getting wrong is the voice instructions at much more important and real junctions. These errors take several forms:

Roundabouts: For instance, entering a five-exit roundabout at six o’clock to leave at exit 4 at, say, eight o’clock. On a French roundabout this requires travel around the roundabout anti-clockwise to the forth exit at eight o’clock. This correct route around the roundabout is clearly shown on the device’s map screen and on the ‘pop-up map’, so the route is correctly entered and the device's screen is showing it correctly.

The voice instruction however says, quite clearly, ‘Leave the roundabout by the first exit’. In my French example, this would be three exits early and be at roughly three o’clock. I can only assume that somehow the device and voice are becoming confused as to which way one goes around a French roundabout. For instance, if it were a UK roundabout then the fourth exit would indeed be the first exit.

When I ignored the voice and followed the map (and road signs) around the rounabout the voice prompt quickly corrected itself. When I sailed past the exit the voice had told me to take it went on to say, ‘Leave the roundabout by the XX exit’, almost as if the voice had now realised its error and caught up with the map.

Distinct junctions or distinct deviations from one road onto another: Here I am not talking about new roads that do not (for understandable reasons) appear on the map. Rather, I am talking about long established roads, requiring, say, a right angle turn onto a fresh road or, in the case of a motorway splitting into two, a similar real deviation in the route. Here is an example:

Straight road on the map running north – south. The straight road is the A5. The plotted route and map clearly show the need to leave the A5 left, at right angles and join the B4. In other words, a significant deviation onto a completely fresh road. The voice prompt completely ignored any prompt to turn left. This means you sail straight past the junction and are only made aware the error when the voice pipes up to say, ‘Off route, recalulating…..’

The really odd thing, in one real life example of this happening, was that the voice prompt warned of the need to, ‘Keep left in two miles'. This was not an instruction to turn left in about 100 yards onto the B4, but warning of a sharp bend (see example above) on the B4 itself. Meaning that the voice instruction did not give a basic instruction to turn left onto the B4 but did warn of a bend coming up in two miles on the B4, once the turn had been made and the driver having travelled more than two miles.

A road crossing a motorway: Again this was in France. A country road crosses a motorway at right angles. The country road runs east to west and the motorway north to south. The country road ends in a Y junction. The right hand fork leads down to the motorway ie. it’s a 'ramp' road down to the motorway. The left-hand fork ‘ramps’ left and continues as a country road which proceeds to bend around to run parallel to the motorway.

The route and the map clearly (and correctly) show that the vehicle is to take the left-hand fork, down the 'ramp' continuing on the country road and to run parallel to the motorway. The voice instruction however clearly says, ‘Keep right in three hundred yards (when it should be, ‘Keep left’) and then says ‘Take ramp right’ (onto the motorway) when it should be. ‘Take ramp left….and follow the road…..’, thereby matching the map and downloaded route. It obviously follows that, had I simply followed the voice, I would have ended up on a motorway, heading God knows where, with a voice bleating, 'Off route, recalculating....'

As I said:

(1) All of these examples (and numerous others) are real life and happened over a four-day 2,000 mile trip around northern France. All of us using Garmin devices were equally troubled. Similar things have happened quite regularly in the UK and to me in Spain.

(2) The frustrating thing is that 99% of the rest of the package is very reliable, with only the occasional glitches. These occasional glitches are mostly caused by satellite reception problems, which are hardly all Garmin’s fault.

(3) The real gripe is where the voice instructions are so wrong, or missing or deviating entirely from the map and from the route installed. This leads to a feeling of unreliability and real concerns that any minute another error will be made. This feeling stays with rider and distracts from the job in hand ie. staying upright and avoiding the million and one other hazards set in the way.

(4) If Garmin are interested, I can map out and send examples where the routes went wrong. I guess that Garmin would be able to play them back with a voice prompt running to see what I mean.

Kind regards,

Richard Nash (Wapping)
 
When I came back down the A1 yesterday the Quest v8 abbruptly told me to turn off at the roundabout ajoining the A6. Very strange.
 
Thanks for the information Richard. I'm heading to France in three weeks with the family, by car, so I'll look out for the glitches. I shall be returning with the bike in September fully aware! Currently use a 2610 but hope to go for a 2720 soon.

Andy
 
Interesting. Up to now I haven't had these problems, all the roundabouts and junctions OK except for the usual stuff about turning left or right when it should say bear whatever. V8 has had a good workout between Toulouse and the Pyrenees and in the Cevennes/ Gorges du Tarn area, no major problems at all. You had the problems in Northern France, maybe the proximity to the UK confuses it :D :D
 
Here is an example of the error at a roundabout.

I have used the original routing files, simply cutting and pasting the screen print.

ErrorinV8.jpg


The saved route was to drive south from Calais and leave the A26 motorway at junction 5 to pick up the D916 south. Instructions 10 through 15 below.

V8error3.jpg


The device gave the correct instructions at junction 5. Routing instruction 10.

It then gave the correct instructions to curve around, over the motorway and drive north to the roundabout. Routing instructions 11, 12 and 13.

On approaching the roundabout, the device's map and the pop-up junction map both clearly showed the correct anti-clockwise route around the roundabout.

ErrorinV82.jpg


But the voice instruction instead of saying, 'Leave the roundabout at the fourth exit' (the D916 south) said, ‘Leave the roundabout at the first exit’. This is clearly wrong on two counts:

(1) The correct exit is the fourth exit, clearly shown as the D916 in a southerly direction - matching the routing intructions 14 and 15 shown above.

(2) The first exit is the D69, direction east.
 
Hi Richard:

Thanks for posting such a carefully written account of the difficulties that you have encountered.

I can't think of anything to say, other than that I have brought this post to the attention of the software engineers who (I think) will be interested in it.

By any chance (this is a shot in the dark), were you navigating a route that included travel in both the UK (driving on the left) and the continent (driving on the right)? I'm just wondering if perhaps the transition from LHD to RHD midway along the route might have something to do with it...

I have occasionally noticed that the guidance through a roundabout will skip a single exit from the roundabout, if that exit does not lead to a roadway - for example, if it leads to a private lot, or a roadworks maintenance area, or (I had this happen once in Cambridge) if the exit leads to a bus bay. However, it doesn't sound like that is what you are encountering - your writeup made it pretty clear that the voice guidance was skipping quite a few exit points.

By any chance did you observe what your position on the map was shown as (with the vehicle position symbol) just as you entered the roundabout? What I'm getting at here is this: If the cartography does not depict the roundabout and the roads leading to it in exactly the correct place on the surface of the earth, then the StreetPilot might have compared your position (as derived from the satellite constellation) with what roads the cartography suggested were closest to you, and then 'placed' you on the wrong road as you entered the roundabout. That might (perhaps) be the answer to the problem - in other words, it might have been caused by a cartographic inaccuracy, rather than by an error in the software that issues the voice and visual navigation commands. If you still have the track log for that problem roundabout, don't throw it away...

Respecting the pictures that you posted (just above), can you provide a rough latitude and longitude for that roundabout? Otherwise, the software engineers won't be able to find it, for the purpose of reproducing the problem in Kansas.

Thanks,

Michael
 
The missing left turn error

Here is the left turn error, again simply lifted out of the original Garmin route files.

The routing instructions 45 and 46 correctly show a south-westerly route on the D33

Instruction 47 is the routing instruction to join the D18 and drive south. This too is correct.

Instruction 48 is, I believe, the kink in the D18 left, just after leaving the small village of Pont Douche. I believe this is the manoeuvre the voice instruction warned of whilst we were still on the D33, about two miles before.

leftturnerror2.jpg


The map shows the sharp right angle turn left, from the D33 to the D18. Both are obviously completely separate roads, with different road numbers, travelling in totally different directions.

Leftturnerror1.jpg


No voice instruction was received to turn left. The first indication of a mistake was an, ‘Off route….recalculating’ voice message as we sailed past the junction and started to head under the impressive bridge carrying the motorway.
 
Wapping said:
Here is an example: Straight road on the map running north – south. The straight road is the A5. The plotted route and map clearly show the need to leave the A5 left, at right angles and join the B4. In other words, a significant deviation onto a completely fresh road. The voice prompt completely ignored any prompt to turn left. This means you sail straight past the junction and are only made aware the error when the voice pipes up to say, ‘Off route, recalulating…..’

<snip>

...A road crossing a motorway: Again this was in France. A country road crosses a motorway at right angles. The country road runs east to west and the motorway north to south. The country road ends in a Y junction. The right hand fork leads down to the motorway ie. it’s a 'ramp' road down to the motorway. The left-hand fork ‘ramps’ left and continues as a country road which proceeds to bend around to run parallel to the motorway.

The route and the map clearly (and correctly) show that the vehicle is to take the left-hand fork, down the 'ramp' continuing on the country road and to run parallel to the motorway. The voice instruction however clearly says, ‘Keep right in three hundred yards (when it should be, ‘Keep left’) and then says ‘Take ramp right’ (onto the motorway) when it should be. ‘Take ramp left….and follow the road…..’, thereby matching the map and downloaded route. It obviously follows that, had I simply followed the voice, I would have ended up on a motorway, heading God knows where, with a voice bleating, 'Off route, recalculating....'

Hi Richard:

Again, may I trouble you to provide latitude / longitude locations for those two problem points, to allow the software engineers to attempt to reproduce the problem?

Even better: If you could create a MapSource file (a .gdb file) that contains the starting and ending points of the three locations where you encountered the problem, then attach it to this post, that would be just first-class. Here's how to do it:

1) For each problem area, establish a waypoint called "Start A" and "End A" (Start B and End B for the second problem, Start C and End C for the third problem, etc.). Start should be about half a mile before the problem, and end about half a mile after the problem. Create a waypoint called "Problem A" right on top of the problem, to clearly identify the intersection that is the problem one.

2) Create a simple route from Start A to End A. Do not include the 'Problem' waypoint in the route. Save the routes with a unique individual name (Problem Roundabout, Problem Fork, whatever).

3) Now, upload that .gdb file - which will contain all your problems - to your post, by changing the file extension from .gdb to .zip (this because the forum software will not allow an upload of a .gdb file), then using the 'manage attachments' button that is about 3 inches below the box that you use to compose new text for a message (about 2 inches below all the post icons).

Thanks,

Michael
 
PanEuropean said:
Hi Richard:

3) Now, upload that .gdb file - which will contain all your problems - to your post, by changing the file extension from .gdb to .zip (this because the forum software will not allow an upload of a .gdb file)

Michael

It can be uploaded as .gdb :thumb

"Valid file extensions: bmp doc gdb gif jpe jpeg jpg kml kmz pdf png psd txt zip."
 
In fact, here's an even better idea:

Create the three "problem" routes, as described above. Save the MapSource file, then upload those three routes to your GPSR. Put the GPSR in simulation (indoor) mode, then activate the routes one at a time, and observe what the screen looks like - meaning, the pop-up 'next turn' guidance screens, and listen to what navigation instructions are issued. (You will probably have to install the GPSR on your moto and run it in simulation mode there, because you won't have a speaker attached if you just plug it into the wall).

If the GPSR gives you appropriate visual and verbal instructions when running in simulation mode, then the problem is likely with the cartography - the roads are not depicted (in the cartographic database) as being on exactly the same lat/long that they occupy in real life. As a result, the GPSR is placing your vehicle in the wrong position, and that's why you are getting the inappropriate navigation guidance.

But... if the inappropriate results can be duplicated when running in simulation mode - you have found a genuine problem.

Michael
 
PanEuropean said:
Hi Richard:

Again, may I trouble you to provide latitude / longitude locations for those two problem points, to allow the software engineers to attempt to reproduce the problem?

Michael

Hi Mike,

The coordinates of the roundabout error are:

N50 33.453 E2 27.812

This is the entry point to the roundabout.



The coordinates of the left turn error are:

N47 09.987 E4 42.044

This is right on top of the junction.

Richard
 
Mike,

I thought it might be a simple ‘positioning’ error but how likely is it that four motorbikes had exactly the same problem each and every time? For example, the roundabout error example at junction 5 has occurred four times to four separate bikes on different days.

I have several examples of T junctions missing voice instructions. These are not simple deviations in the road but rather ‘proper’ major road intersections where we would be travelling up the leg of the T on a main road, come to the junction, the map showing turn right (the positional arrow showing the bike’s position correctly on the map) onto another major and total silence from the voice instruction. I go to France roughly six times a year and often ride routes which, at some point, cross the same problematic junction(s) – each time, total silence.

The ‘positional’ error theory doesn’t seem to hold water in the left turn example. Here the voice was giving a ‘Keep left’ instruction about two miles before the turn off left ie. flagging up something it ‘knew about' two miles away on a road we would be turning onto but was silent about the actual left turn itself. Surely, if it could tell where we were two miles before the slight left deviation it should have spotted the junction only 200 yards or so away.

Nor does it seem to stand-up in the roundabout problem.

Somewhat sadly I have other examples of roundabouts where the ‘Take the first exit’ voice instruction pipes up in error for an exit right around the other side. If it’s a positioning error, where can the device possibly think I am in relation to the exit? In France, I would have to be at roughly 10 o’clock to my actual position and heading in the opposite direction, surely?

To answer your two questions:

(1) By any chance (this is a shot in the dark), were you navigating a route that included travel in both the UK (driving on the left) and the continent (driving on the right)? I'm just wondering if perhaps the transition from LHD to RHD midway along the route might have something to do with it...

No. I have totally separate UK and Continental routes and never one that crosses the gap between Dover and Calais. I also try to break routes up into manageable portions, rather than having several days travel on one route. If nothing else it saves recalculation time.

(2) I have occasionally noticed that the guidance through a roundabout will skip a single exit from the roundabout, if that exit does not lead to a roadway - for example, if it leads to a private lot, or a roadworks maintenance area, or (I had this happen once in Cambridge) if the exit leads to a bus bay. However, it doesn't sound like that is what you are encountering - your writeup made it pretty clear that the voice guidance was skipping quite a few exit points.

I agree that the device will sometimes skip a ‘dead-end’ exit. Sometimes it will also give credit to an exit that it is not an exit all. These small glitches I can live with, it’s only a guide after all and the road signs / map give a pretty good clue. But neither was case with the example I gave. As you deduced yourself, this was a major motorway feeder road type roundabout of significant dimensions and with sizable exit routes around the clockface.

I thought about running the simulation on my GPS device and will do so. As you observe, this involves mounting the device to my bike and using the speakers in my helmet via the Autocom. I will do it at the weekend in the light.

Richard
 
Hi Richard:

A follow-up question: What voice were you using? American English, British English, French, etc? Also, what version of the voice? You can determine the version number of the voice software by checking somewhere in the GPSR... I can't remember exactly how to do this, and I am in Africa now, without any automotive GPSRs... but I am pretty sure you will figure it out. The voice version information is in the same place as the software version information.

Many thanks,

Michael
 
PanEuropean said:
Hi Richard:A follow-up question: What voice were you using? American English, British English, French, etc? Also, what version of the voice? You can determine the version number of the voice software by checking somewhere in the GPSR... I can't remember exactly how to do this, and I am in Africa now, without any automotive GPSRs... but I am pretty sure you will figure it out. The voice version information is in the same place as the software version information.Many thanks,Michael

Hi Mike,

Thanks for the interest you are taking.

If you are in Africa you may well be without a lot of things, let alone GPRS :)

Software version: 4.60

Audio version: 2.10

Voice: Guidance and Status

Text Language: English

Voice Language: British English

Guidance: Off-Route Recaclulation: Automatic - Announced

Next Turn Pop-up: Both

Safety: Safe mode: Off
 
I'm using CN8 with my newly replaced Quest 2 and the very first thing I noticed was that my village now has a sub post office from a village that is 10 miles away, and we've also got a new pub (although it closed down over 10 years ago). These errors didn't appear in either CS6 or CS7.

Progress ???
 
I'm running V8 and have have used both it's predessor versions in a Nav II with the voice input.

I put the audio variation down to US vernacular, "keep left ahead" as meaning stay on this road. "Turn right" as stay on the main road when the road deviates and there is a nearby junction when we would describe that as 'bear right' or 'just stay on the major road'.

This has been the same in all three versions that I've used.

It is off-putting when you look at the visual prompt and in "English" you think, aha, I just stay on the main road, while in "American" you get an instruction to 'bear left' or whatever.

And why hasn't she got a French accent in France? (I'd forgive her anything then! :D )
 
MikeP said:
I'm running V8 and have have used both it's predessor versions in a Nav II with the voice input.I put the audio variation down to US vernacular, "keep left ahead" as meaning stay on this road. "Turn right" as stay on the main road when the road deviates and there is a nearby junction when we would describe that as 'bear right' or 'just stay on the major road'.
This has been the same in all three versions that I've used.
It is off-putting when you look at the visual prompt and in "English" you think, aha, I just stay on the main road, while in "American" you get an instruction to 'bear left' or whatever.
And why hasn't she got a French accent in France? (I'd forgive her anything then! :D )

Mike,

Thanks for your thoughts. Maybe a cute cute French accent would indeed help :)

I can cope with the vagary of the language used. That is not the problem.

What I seem to be suffering from is a complete absence of voice instruction at some major junctions and / or verbal instructions that are completely wrong ('Keep right and take slip road right', when the direction on the map and the road is clearly, 'Keep left and take slip road left') in relation to the map and the plotted route. The roundabout and Y fork examples show this well. Even allowing for langauge differences, muddling left and right and exit one from exit four is difficult to explain.

The only way that I see the voice instruction muddling left and right and misquoting the exit from a roundabout is if it is somehow reading the direction of travel as reversed and maybe calculating the vehicle as being on the other side of the road. If this were the case then right would indeed become left and the fourth exit of a roundabout becomes the first. But this is not what the map screen is showing, so it seems unlikely, unless the voice is somehow de-linked from the map – equally unlikely I guess. However, even if this were the case it would not explain the complete absence of voice commands at some major junctions.

On my last jaunt to France we had three bikes all running Garmin V8 on Garmin (or Garmin variant GPS devices) and were also linked bike-to-bike by radio. There were about eight other bikes in the party so the three of us were spread over sometimes a quarter of a mile on the road. Whenever the lead navigation bike had a glitch it was mirrored down through the other two bikes. All three of us are regular navigation users so are used to the foibles, so it is hard to put it down to user error. The lead bike got used to transmitting, ‘She misses out the turn left / right’ or ‘She means the fourth exit, not the first’ and asking the others to see if they too suffered the glitch. With 100% certainty the other two bikes encountered the same error.

Richard
 
Hi Richard:

Thanks again for the very comprehensive information you provided (at the bottom of page 1) about what software, voice, etc. versions you are using. The software engineers are following this thread with considerable interest, and started looking into the matter yesterday.

Michael
 


Back
Top Bottom