Inconsistent Route Calculation Garmin 52LM

What avoidances are set on the 52M? Tolls, U-turns, etc.

So, a shorter distance equals a faster time in a lot of instances. It doesn't know anything about stop lights, stop signs, traffic congestion or other impediments so if the calculated speed over the surface street is close to the time it would take to do the logical route, imbecility wins.

Checked them even though they did not seem to apply since I am simply adding 2 miles of interstate between the 2 starting points. They were U tuens, carpool lanes and unpaved roads.

Also should point out that I did not do a map update because the docs say a high speed internet connection is required. I only have a SLOW speed DSL connection. But, having said this the maps are correct. So the update would seemingly need to be an OP system update also.

I do my updates on a 3mb DSL connection. It's fast enough.

...but having difficulty following the math. Meaning, if it takes 2 minutes to get from the second starting point to the bridge, and A mins to get from the bridge to Sonic via exit 157a, vs. B mins via 162, then how can A be less than B, and (A + 2) be greater than (B + 2)?

Phranc, thanks. Good to know! Do you know if the update is more than a map update? (Nonetheless, I still may have a problem because my DSL is only 750 kb.)

If you have a laptop, take it over to a McDonalds or Starbucks and use their WiFi connection. My experience is you will have fewer problems at Mickey's as there are fewer people there trying to use the WiFi. (That and a senior coffee is cheap and free refills.)

Understand that the "route time" is calculated using the programmed "average speed" of the road types on the route.

It does not take into account traffic lights, stop signs, actual speed on the road, etc.

Software updates are very much smaller than map updates. Map updates run 4-5gb and software updates are usually only a few megabytes.

But doing a map update over a 750kb line is going to be very painful. I think the last update I did one on a 3mps dsl line took something like 8 hours.

I think I've experienced at times and with some map/software versions that some nuvis will try to avoid making certain left turns and will create a route that's longer to avoid the turn. Could this explain your findings?

Does either route involve any areas that have been altered with new roads, intersections, speed limits, etc. in maybe the past few years? If so, a map update now or some future version may correct things.

Also, over time and as long as you don't do a hard reset, the nuvi will learn the speeds you travel at for the various types of roads and speed limits. It could be that your new 52 will eventually take the route you know to be faster.

You also asked about software updates. The latest software update that I think matches your 52 is version 4.00 from last August:

http://www8.garmin.com/support/download_details.jsp?id=5937

The update is less than 11MB so you could try updating your device software using Garmin WebUpdater if your version is an earlier one. It will download much faster than a full map and software update.

https://support.garmin.com/support/searchSupport/case.faces?...

One last thought: when in familiar areas where you know a better route than the one offered, take the route you know to be faster and let the nuvi recalculate.

As Box Car points out, nuvis can make bad guesses. But when you're in an area new to you, arriving at the desired destination, even if not by the desired fastest or shortest route, can still be considered a success less than perfect but still a success.

Good luck and have fun with it.

Just ran the above 2 simulations on the older C530. Both simulations gave the same result...take the 162 exit on Bluebonnet. ....which is the right answer.

CraigW, thanks for the software update suggestions! I'll log in and check this out. ... I'm not worried about having poor advice on routes where I know better. I am more concerned about poor advice on routes where I don't know better...In any case, I would hope the advice to be at least as good as with the older C530!

Also, had lots of "fun" with the C530. Expect no less with the newer device!

After my last reply, I thought of two more questions for you.

The current City Navigator map version is 2015.30. What version of map do you have on your 52 and on your C530? The older they are and the bigger the difference in version between your two devices, the more likely it is that there could be a routing error or difference in routing between devices.

When you do arrive at your destination, was the destination selected from the Food category choosing Sonic or did you select the destination as the street address? Whichever method you used, does the Sonic appear along Perkins Rd at essentially the proper location and on the correct side of the street? If not, a map update may correct the location in the new map's built-in POI or street address database.

CraigW, I did verify that my map software is NOT the August 4.0 version. So I will be updating this. Senondly, I tried multiple destinations in the nearby area using street addresses and still got the inconsistency.

My guess is that this is a Garmin software problem and not a map problem because:

1) the routes from the bridge to the one and only destination are identical. The only difference came by adding 2 miles of interstate at the beginning. Any mapping problems would be duplicated in both simulations...the.same is true for speed limits, stop signs, traffic lights, left turns, etc.

2) the older Garmin C530 gets the same (right) answer. The added 2 miles of interstate at the beginning does not result in a different route determination.

By the way, you folks can all run the same simulations. All of the information needed is stated above. If you do, I'd be interested in hearing your results, particularly from folks with the 52LM with the updated software.

I tried both routes as you described. Each time I was directed off I-10 at exit 157. At least mine is consistent if not the best route.

By the way I have 2015.30 maps, Version 3.50, faster time no ferries car pool lanes unpaved roads.

Getting off at Exit 157 results in a route that is shorter by over a mile and according to the algorithms Garmin uses where it does not consider any stop lights or signs it should be faster. In the years since the C530 was produced and the Nuvi line came out, there have been many programming changes to the software embedded in the units. Not picking routes based on the class of road appears to be one that changed.

I have an older nuvi 250 and I catch it doing this kind of thing all of the time.

I only use "fastest" when traveling, locally I have it on "shortest". Yet it consistently passes up shorter routes and tries to put me on longer router with more traffic and more traffic lights. It does know all of the data for the shorter route though, since as soon as I turn on to it and force it do do it's "recalculating" it will take that route and the ETA drops and the remaining distance drops. The route isn't being avoided for any reason either, as if I approach that road from a different direction Garmin will tell me to take it (rather than tell me to make a U-turn and go the slower longer route).

I have also seen completely different routes suggested with no good reason. The above happens when I'm driving home northbound at about what I'll call point A on road A'. I have also told the GPS to route me home from a street B' that ends at street A' a little south of point A. Any suggested route from point B would go through point A. So I would expect that it would get me to point A (a very obvious path with no real alternative) and the suggest the same route the rest of the way home. But it has come up with alternate routes home, taking turns off the route from A to home. And it even picks two different alternate routes, which I think happen for slightly different starting locations, depending exactly where I am on road B'. (the roads suggester are better than the suggester route from A to home but not better (IMHO) than the route that Garmin continues to ignore from A to home until I force the "recalculating".

Recently (with the GPS set to fastest) I had it calculate a back roads trip home. The route was workable, but there were clearly a few serious poor selections. Multiple times it had me routed on two roads from point C to D (a small subset of the trip) when points C and D were on the same road. When I spotted this I just took the direct road, but the Garmin even tried to get me to take a U-turn and correct my route, it didn't shut up until we were a ways past point D and it finally realized I wasn't going to go back.

Of course, these are the problems that I know about, either because I know the area and know the choice being offered is poor or even because I can see where I'm trying to get to an am being routed a clearly slower longer way. But when using the GPS for travel in a new area I don't always know if the choices made are right. The GPS will usually end up getting me there, but if it is making poor choices when I know the area then it is likely making poor choices when I don't know the area and have to take its word on the route.

In fact, last year it took me on a surprise route, at first down a major highway, but then that ended and I was on a long tour of the back roads of Kentucky. I could have avoided that if I had checked the route earlier (actually I did but adding a way-point that was along the original desirable route was enough for Garmin to decide to recalculate the route in its new creative back-roads way! And sadly I didn't recheck after adding the simple way-point while driving. The way-point had only been added to see how far I was from the cheapest gas on the route so I would know how much gas to get at a more expensive stop).

OK, you've shamed me into looking at a map

Using Mapsource, I got a good view of your two routes. Ah, I see your concern.

I then fired up a nuvi 3597 with latest firmware and 2015.30 map and "standard" avoidances. For me, whether starting on the east side of the I-10 eastbound bridge over the river or the west side of the bridge, my nuvi chooses Exit 162a exit in Fastest mode.

I then tested a nuvi 2460 with the same setup as above. I only tested a start from the east side of the bridge as that is where your routing problem exists. I again was Fastest Routed to Exit 162a.

I then tested a nuvi 855 with the same setup as the 2460. Fastest again took me to 162a.

Finally, I charged the battery and fired up a nuvi 295W with the same setup as the 2460 except that the map was an ancient 2011.40 map. From the east side of the bridge, my fastest route to Sonic took exit 160, not quite your 157a but also not 162a!

So from my testing, using nuvis with the latest firmware and latest maps, I am consistently being sent along the route you say is the fastest, using exit 162a. But using an ancient map on a nuvi 295W, I'm sent along a slower route, close to the one your 52LM is sending you along.

My suggestion therefore, is to update your firmware and map, then repeat your test. It's certainly possible that you will be sent along the same (correct) fastest route regardless of which side of the bridge you start from. An alternative answer, one we don't want to hear, is that your nuvi 52 and my nuvi 295W have a less accurate routing algorithm which will incorrectly choose a route in your example that is not the fastest route. As you mention, we really would like to read another nuvi 52 user's results using the latest firmware and maps to run your two simulations and see what route is selected. mldorsey's results with a nuvi 50 may suggest that the Essential Series of nuvi does have a less accurate routing algorithm than the various "premium" nuvis have.

You mention your slow DSL connection at home. If you have a laptop, what about going to a library or McDonalds to use their wifi, then do the map update there? Whatever connection you use, know that a well over 2GB download is going to take some time. Make sure that the computer is not set to go to sleep during the map download/update process.

Hmm, we need a Garmin Express person now. If Garmin Express is installed and allowed to run in background, won't it download the map update in background even if it takes a day or more? If so, that could be an alternative way to download the latest map while sticking with your DSL.

It's supposed to. That seems to be the sole redeeming value of running GE in the background. I would start it up and then close it and let it do it's thing.

You make an assertion and that assertion is that the Fastest Time route is using Bluebonnet Blvd. Now, I realize that this is your area and you likely have experience that this particular route is what has proven to be the fastest for you.

As usual, Box Car has given us the essential elements to use when discussion your findings.

However, consider what your nuvi has to work with.

It DOES NOT have all of the traffic lights and stop signs in the country in its database (and we would not want it to have because we already complain about the lack of space on the units). It does know types of roads and speed limits. It tries to do the best it can with the data it has.

I tried an experiment with Mapquest to give us an illustration of another companies approach to routing. I started at
Latitude: 30.445939 Longitude: -91.243891 and asked for routes to 10570 Perkins Rd.

The routes were
1. I-10 E to LA-427 12.28 miles
16 mins / 17 mins based on current traffic

2. I-10 E to Bluebonnet Blvd 12.61 miles
16 mins / 16 mins based on current traffic

3. I-10 E 11.43 miles
17 mins / 18 mins based on current traffic

My feeling is that we could try other routing engines and find that exiting at 157A is both shorter as well as faster.

There is not a "bug" in the 52 in my opinion.

Nuvis that speak Garmin Real Directions show that they receive data from the map telling of traffic light locations and maybe stop signs as well, so devices could have access to the data when making routing choices.

Oh no, if 157a is faster, then all my nuvis are wrong

After thinking about it all, the biggest point of confusion I still have is that Grand-Dad's nuvi has varying answers depending on just how far west its starting location is from the first of the various exit options. But from your MapQuest results, it may just be a matter of a couple tens of seconds and therefore is a tossup. The "premium" nuvis that offer trafficTrends could easily change which I-10 exit it suggests depending upon time of day and day of week, but I don't think the 52 offers those trends. I think all/most nuvis do learn its drivers' speed vs. speed limit (with all unlearned after a hard reset), so that over time, one person's nuvi may suggest an early interstate exit while another's a later exit. One final complication is that when taking exit 157a, the route passes two schools and for nuvis that offer school zone speed changes, the exit 157a route could have two slow downs to maybe 15 mph during school days/hours. This may be especially important given its location when we think back to when our Factory member rame1012 shocked us all pointing out that some school zones in Louisiana had time of day speed limit changes built into the City Navigator map that affected all nuvis, if I remember correctly. Goodness, just look at what you've created, Grand-Dad!

The great news is that all nuvis do get their drivers to the proper destination for a burger or hot dog.

My 3490 sometimes takes me a different route even though all setting remain unchanged.

Weird.

CraigW, updated to 4.0 per your recommendation. Still getting different results for the 2 starting points to the above Perkins Rd destination but the results are now the reverse of what fhey wrre before. I then decided to run 2 more simulations, one with the starting point in the city of Lafayette (50+ miles west of the bridge), the other about half way from Lafayette to the bridge. The results are to take exit 162A if you start in Lafayette or on the I-10 Mississippi river bridge, however you would take exit 157A if you start at points in between.

So I decided to call Garmin Tech Support. I explained the problem and pointed them to this thread. They suggested first updating the map which would take about 8 hrs on my slow DSL connection. (Now in progress.) They also said that they would look into this issue. If all else fails I can submit this as a mapping issue whereupon they would research this and make any changes if deemed nevessary.

Hey Grandad-2003. I spoke to you earlier and have tested on my nuvi 44LM with latest mapping, software, and identical navigation settings. Just for reference, the nuvi 40, 42, 44, 50, 52, 54 are the same series and use the same exact software file (135000010400.rgn).

I set a starting location of 2852 Charles Dr, Port Allen, LA. This is a random address a few miles west of Baton Rouge, along I-10.

The destination is set as Sonic Drive-In (4171 Perkins Rd).

The device directs me onto I-10 E and then taking exit 157A to Perkins Rd.

***

I then moved the starting location a few miles further West, to 76735 Garner Ln, Grosse Tete, LA.

Entered the same destination of Sonic Drive-In on Perkins Rd.

I am still being directed to stay on I-10E and take Exit 157A.

Getting the exact same results with Advanced series devices using Garmin Guidance 2.0 and 3.0 as well.

***

As advised when speaking with you, please update your mapping from 2014.10 (current is 2015.30). The issue you are experiencing is likely due to an error in the older mapping.

If you still encounter any issues, please give me coordinates or starting addresses and we can certainly test further.

***

Lastly, I do want to mention a bit on how the device calculates ETA for routing in general.

The three main points are:

Speed Information (including real-time or historical traffic information)

Turn costs (turn angles, speed categories of roads, etc)

Traversal costs (likelihood of having to wait or slow down for intersections)

Most devices also do learn from drivers and will adjust ETA based on user history (say you always go 10 miles over an interstate, but follow the posted speed on city roads; the device will begin to prefer the interstate route even when the city roads would have been slightly faster).

NothingOfficial, yes, Exit 157a is going to be the correct exit to take from I-10 to reach the Sonic at 4171. But Grand-Dad's looking for directions to the Sonic at 10570 Perkins Road. Try that destination and see what I-10 exit you find.

Exit 157A is the correct exit for that Sonic. The address of the Sonic I used is 10570 Perkins Rd

Ah, my mistake. He merely stated Sonic on Perkins Rd when speaking with him; I should have double-checked the address here.

I am getting the same results listed in the original scenario on an updated device.

Please note, this does not mean that there is a bug in the routing algorithm. As stated previously, the algorithm takes into account quite a few factors and generally tries to keep the user on similar road types.

If you are on a long highway route, your traversal cost to take an earlier exit will be higher, as there is greater chance to run into a slow down.

If you start closer to the destination though, the same exit/road will also calculate a lower turn cost.

Overall, there is only a slight difference in ETA between each exit. If you feel it is in error, Grandad-2003, please feel free to submit the map error report as previously discussed:

http://support.garmin.com/support/searchSupport/simpleCase.htm?caseId={3a6009f0-44ee-11dc-4733-000000000000}&kbName=garmin

I drove to work every day from an address near the Sonic towards the Mississippi river bridge (meaning the opposite direction from above). I timed the drive multiple times. On average the interstate route (162A route) is about 2 to 3 minutes faster. It is slower when there is an accident on the interstate. Also, the 157A route is usually faster at peak PM rush hour traffic ...around 5 PM.

The map load is a little over half complete after 5 hours...so still in progress.

The map load is complete... not sure how long it took because I left it to complete overnight... My guess would be around 10 hours. FWIW

The results on the cases described above are now giving the same answer... take exit 157A.

As noted by posters above one could build a case for the map data yielding a result whereupon fastest time is also the shortest distance. Having lived in the city for almost 39 years, I believe that on the average, the 162A is the faster route to take. Nonetheless, there is logic to the getting this result - again, logic due to the map data.

So I decided to run 2 more simulations just for kicks using locations near parents/in-laws. Both of these are in the vicinity of Lafayette, LA.. which is about 55 miles west of the I-10 Mississippi river bridge. One starting location is on the northwest side of Lafayette (N30 14.354, W092 06.446), the other starting location is on the southeast side of Lafayette (N30 11.273, W092 00.893). The destination is as before, 10570 Perkins Rd, Baton Rouge, LA. The results are to take 157A in Baton Rouge if driving from northeast of Lafayette, however take 162A in Baton Rouge if driving from the southeast side of Lafayette.

Something just doesn't jive with this. How can a mapping program (and/or map data) give 2 different driving routes in the destination city which is 57 miles from the starting points? I just can't seem to believe this is an acceptable result. If I had developed the algorithm and/or programming for the route determination, I'd have to look into this anomaly...but I guess that's just my opinion. So I will submit a map error report as was recommended above.

Hey folks. I really appreciate all of the good feedback from everyone. Thanks!!!

It is not at all surprising to me that you might get a different routing result if you change the starting point, even if the destination segment of the route is the same. The reason is because the routing algorithm uses "fuzzy logic" which means it does not evaluate every possible combination of roads, turns, etc between a starting point and destination. The underpowered processor in a nuvi would take forever to try the millions of combinations. So when you change the starting point, you change the number of possible decision points, and therefore the number of routes that can be tested at each decision point.

In my working days, I did some fuzzy logic programming (student scheduling application). That type of program is a completely different animal than something like an accounting application that uses fixed formulas and algorithms.

FWIW, I simulated both routes from Lafayette on my 2597, latest maps, latest updates, and both routes had me exiting at 162A.

On this, as long as you get to your destination,that is what counts.

Grand-Dad, my guess is that you are a likely candidate to not perform each quarterly map update. There's certainly nothing wrong with that.

This is a reminder, though, that one of Garmin's terms&conditions about a year or so ago changed and now states that devices with lifetime maps will continue to receive map updates so long as map updates are performed at least every 24 months. Going more than 24 months after a map update loses the device's guaranteed ability to receive future updates. I think this new policy is less than 24 months old so it's not clear how rigidly 2-yr unupdated devices will lose their lifetime map update ability.

Should you look to buy a new Garmin nuvi in the future, you might want to look at a model above the "Essential" series like your 52LM and look for one that offers trafficTrends since you stated that during Rush Hour, you think exit 157a is faster than staying longer on the interstate while you feel staying on I-10 longer is faster the rest of the time:

https://support.garmin.com/support/searchSupport/case.faces?...

https://www.youtube.com/watch?v=WUxnPrqqchA

myTrends is another possible plus for a future purchase:

https://www.youtube.com/watch?v=lJQdB8Eahbs

Alandb, if fuzzy logic is what Garmin uses then that helps in explaining how the result could be different. But, what about comment from phranc?

Phranc, interesting. See comment from NothingOfficial" above. I guess the software in your model is different and not compatible with the Nuvi models mentioned above. ....Because your software gives what I believe to be the right answer.

CraigW, it's not about the traffic. I only mentioned that as a way to partially defend the 157A exit solution. I did not want one of the traffic monitoring models. Instead, my issue was my 7 year old C530 gives a consistent and (IMO) better answer than the 52.

There can also be a difference, as noted by NothingOfficial, that user's driving habits can affect results. Phranc's and my results may differ after we perform a hard reset of our devices. The only way to truly compare different models is to test each immediately after a device hard reset. But for the user owning the device, the longer the device is driven after a hard reset, the more accurate the results will be for its driver(s). I expect that phranc's and my devices have much more user results than your recently purchased 52LM and that our three driving habits may alter results in any event.

All in all, rthibodaux's comment may be the latest of the final conclusion, especially since the difference in fastest route chosen may only be one minute. I don't see the issue in this thread as being a defect in routing.

Good point. I just this AM mentioned to my wife that we have not yet driven down Perkins Rd. So the Nuvi 52 does not know how slow that route actually is.

Rthibodeaux was not the first to make that comment...but using that logic, Garmin would not have allowed for avoidances or put in a system that monitors traffic. Also, fastest time would need not need be a selection. ...in the end everyone is free to choose their own routing preferences. (...And, on the average the difference in travel time for 162A vs 157A is more than 1 minute IMO.)

The typical published routing algorithm divides the trip into regions and tries to link good routes through each adjacent region. They need to be able to find a good choice quickly because no one will wait hours (or more) for the “best” choice. (That is not that different than what you would do if taking a multi-state trip and looking at printed maps for each state.) Then large regions may be further subdivided to refine the route. In practice the answer is usually good and quite often optimal.

Since the computer cannot even come close to checking every route, it uses rules to determine when it has “probably” found the “best” route segment. There might be a better route, but unless you are familiar with the area you will rarely be able to tell. Now if you change your starting (or ending) point slightly the regions "may" be divided dramatically different. Especially where there are multiple good options, the selected option could be unexpectedly different.

Demonstrating that the unit was theoretically capable of picking a better route is not that hard. I don’t mean better based on traffic, I mean better based on stored data. Take a trip of some length where there are multiple branching route possibilities that you have some familiarity with. When you know a “better” route option take it and observe the change in estimated arrival time when the GPS recalculates. As expected, the ETA will normally get a few minutes later. But every once in a while the ETA will improve. When it does you just demonstrated that the new route is faster than the route it had previously calculated, thereby showing that the original route was not the best.

Zeaflal, Excellent information.

As mentioned above, I have not yet taken the 157A route with the 52. So there is currently no new/additional route data in our Nuvi 52.

Just wondering ... Is the anonomous data that Garmin is collecting from devices to be incorporated in future updates and used to make better route determinations?

I would say that the data collected is more than likely used to tweak routing algorithms, correct mapping errors, and help develop over all trends in local routing. It's my assumption that just as much local routing is done as is long distance.

Long distance routing is very easily done, stretch a line between Point A and Point B and see what major highway route matches this line.

Local routing takes a lot more horsepower, as you have seen, to develop the fastest routes using a multitude of different road types, traffic trends, and road speeds. This is where actual driving data would do the most good.

The classic and well known routing algorithm is A* (A star). You can read about it in Wikipedia:
http://en.wikipedia.org/wiki/A*_search_algorithm

Almost guaranteed that Garmin uses something far more efficient. But they probably loose some accuracy in order to get that efficiency.

I have noticed that route calculations will often avoid roads that do not have a speed limit included in the map file. If it doesn't know the speed limit, it can't calculate it. +1 on the traffic lights. Time calculations assume that you maintain the speed limit through those intersections, so rely on your experience if you know better.
One exception was a time when my device routed me through a state park in N. GA that was not paved and more suitable for 4-wheelers than highway vehicles. Luckily I was in a pickup truck.