GPS Street Address Accuracy
  • Richkut
    Posts: 44
    Admin Note:
    We have an article titled How to Correct a GPS Map Error if you want to go about fixing issues you have seen.


    I have read dozens of pages about the pros and cons of the NUVI 765T--all understood. However, in all of my reading I have not ONCE ever read a thread concerning the ACCURACY of this GPS!! Since it does not use WAAS, many people that I talk to claim that it could be rather inaccurate. One colleague told me that his Streetpilot 2XXX series gets him right to his driveway, whereas a new NUVI 7XX? that he purchased for his mother only has accuracy to three houses away! From personal experience, my Garmin iQue3600 sometimes gets me to within a house or two (with WAAS), but sometimes only is accurate to a block away!! Now I don't know what is important to all of you (bells, whistles or more useful things such as lane guidance and traffic reports), but shouldn't a GPS be as accurate as possible?!! While my iQue 3600 sometimes notes an accuracy of 10' or less, in reality this is not true (for it is more like 50', which for most situations is good enough unless I was driving blindly on a bridge overlooking the ocean, LOL!!). So folks, what has been your experience not only with the 765T's CLAIMED accuracy (on the xcreen) but with actual accuracy as seen out of your car window? Thanks in advance, as your hoest answers will help me to pare down my next GPS purchase decisions.

    Rich
  • My 760 tends to get me in front of the correct house.
  • Tim
    Posts: 19,394
    The truth is, "accuracy" is pretty much meaningless when it comes to a GPS. Now before you go throwing tomatoes at me for saying that-- let me explain.

    The position your GPS calculates you at... the raw latitude and longitude will be accurate to within 10 meters, 95% of the time. Sounds pretty good, right? And if you were to add WAAS the signal will be accurate to about 5 meters, 95% of the time. Even better, right?

    Well... no. See what you are talking about is the accuracy of the map, not the accuracy of the GPS signal. And the accuracy of the map is nowhere near the accuracy of the GPS signal. When we consider where the street was mapped versus where it actually is it isn't uncommon for the map to be off by 100 feet or so.

    When it comes to the accuracy of a street address, say "500 Main Street" it can be off even further... sometimes by 1,000 feet and sometimes by even more.

    The map will be off to a further amount than the GPS signal will be, so the accuracy of the GPS's calculated position becomes almost irrelevant. 5 meters or 10 meters off will not make any difference in the performance of your street GPS as the weakest link is not the calculated GPS position, but rather than the map.

    So now let's talk about maps. For the GPS devices sold in North America, all use maps from one of two companies, NAVTEQ and Tele Atlas. Both will have neighborhoods that are mapped well, and both will have neighborhoods that are poorly mapped or missed completely.

    So the conversation about "accuracy" almost becomes a mute point. The signal the GPS will calculate is as accurate as it needs to be, based on the weaker accuracy of the map.

    And getting back to WAAS for a second, you might now see why it isn't important at all in a street GPS as the map will be a weaker link. WAAS drains the battery faster and therefore WAAS is disabled in most street based GPS devices since it provides no practical benefit and will just drain the battery faster.
  • Richkut
    Posts: 44
    Thanks for that clarification! However, then, how can you explain the gross differences in accuracy between the above mentioned Garmin units (as they were on the same dashboard of the same car at the same time)? Were the Streetpilot maps more accurate (with or without WAAS) than the newer NUVI maps? I thought that all of the mapsets were the same for Garmin automotive GPS? Would you please elaborate? Again, what "accuracy" have YOU experienced with the 765T? In other words, do you see a flag on your 765T (if you even own that model) right at your own driveway, or is it 5 houses away? I just want to know what to expect in this unit prior to even considering purchasing it, so thanks in advance.

    Rich
  • Tim
    Posts: 19,394
    If all of those devices had the same map version, then they should all report nearly identical results. So I suspect they were not running the same map version if there was a significant (5 houses) difference in the devices.

    In other words, do you see a flag on your 765T right at your own driveway, or is it 5 houses away?

    It boils down like this. For my driveway, NAVTEQ has that street number marked within about 200 feet of where it "should" be... pretty good. Tele Atlas has it about 600 feet from where it "should" be.

    So on every Garmin device I have, as well as any Magellan as they too use NAVTEQ maps, it shows my driveway about 200 feet away when I'm parked there. On any device that uses Tele Atlas maps (TomTom, Mio, etc) my house will show about 600 feet away from where it should be mapped.

    There will be some variance between the devices. Given that one device could be off by roughly 10 meters in one direction and another device could be off by 10 meters in the other direction they could show a difference of about 60 feet from each other, provided they are using the same map vendor and version.

    So even if the GPS was "perfect" at calculating my postion (no error)-- it would still show me as 200 feet away from Home, even on the NAVTEQ map as the map itself is where the greater deficiency is.

    But to get blunt and to the root of the question, I wouldn't have any hesitation going with the 765T over some of the older STP or iQue models with regards to "accuracy". The 756T will have a newer map and therefore hopefully a more accurate map as well.
  • Boyd
    Posts: 11,613
    Also remember that the Nuvi (and just about every automotive unit) apply "road snap" to you actual position. The assumption is that you are driving on a road, so even if you are 50 or more feet away from the road (in your driveway or an unmapped road), it will probably snap you to the road. This may add to any hard error in the GPS fix.

    Truth is, almost any of the modern units will be "good enough" for street navigation.
  • gatorguy
    Posts: 7,244
    Tim, that is absolutely the clearest and most complete set of posts I've seen from any of us on this frequent topic. Practice makes perfect, Huh?😀

    Should make this a sticky and change the topic name.
  • Tim
    Posts: 19,394
    Well, to be honest it is an article I've been meaning to write for a couple of months now, so much of the brainstorming had already been done. I will take it, add some images, provide further explanations, and likely build a article out of it.
  • I can also vouch that is a map issue not necessarily the GPS. My in-laws home is located on end of a street but Google Maps and the GPS's map say it is on the other end of the street.


    |
    |
    |
    |
    GoogleMaps/GPS Map O-------------------------------O Correct Location
  • Richkut
    Posts: 44
    Excellent information and explanation, my friends! This will help my decision-making ever-so-much, and I look forward to the upcoming article/thread on GPS "accuracy"!!

    Rich
  • Barcard
    Posts: 72
    Speaking of map errors, updated maps will correct old errors, but can also introduce new errors. After paying for an updated Teleatlas map for our TomTom, I learned that Teleatlas had moved my house from its correct location to the wrong side of the street.

    On the other hand, our Garmin 755T, using Navteq maps, puts our house right in the middle of the intersection (we have a corner lot) at all zoom levels. However on the route home, Garmin's last turn instruction - both visual and voice - is "turn left on xxxxx street," which is correct, and then announces and shows our home on the left side of the street, which is also correct. So Garmin and/or Navteq may have correctly mapped our home, but choose to display the Home icon in the street. What does that say about GPS accuracy?

    It is true that many addresses will be mapped a few houses off or even several hundred feet off. The solution, if you have the time, is to stand in the driveway, or wherever you feel is appropriate, and mark that location as your home or your destination. Of course you have to be there first. :lol: But if it's somewhere you visit often it could be worth doing.
  • dsrussell
    Posts: 206
    Terrific explanation, Tim. I suspect we all see that same "accuracy" question come up time and time again. I have also noticed on my 765t that as I'm just receiving satellite signal and I check the satellite info screen, it may note the accuracy as large as 50 or 60 feet. But as I gain more satellites, this number decreases, and can even reach close to 15 feet (usually between 17 and 22 feet).

    Another anomaly when looking at the satellite info screen when comparing two 765t's side by side (my girlfriend's and mine), is which satellites are picked up (they are slightly different -- not only which satellites, but how many). I suspect this has to do to where each unit was turned on and working the last time it found signals (we live around 20 miles apart). Looking at them both right now, my unit shows 7 satellites (plus one blinking), and the other shows 5 satellites -- one just barely (plus two blinking). Accuracy is also slightly different.

    Of course, my 765t can also do some strange things. While driving down the freeway, it sometimes "forgets" where I am, or thinks I'm suddenly on an adjacent road, then goes into recalculation mode (it corrects rapidly though). Doesn't happen often, but I took it to mean a combination of map accuracy at that certain area and perhaps temporarily having a minimum of satellites locked on. Where this happens often is at one particular freeway interchange (270 degree loop, where it can recalculate as many as three times before I'm out of the loop).
  • Tim
    Posts: 19,394
    Which brings to mind a couple of other things I am planning to put into the article but will mention here. The "accuracy" reading you see on the satellite info screen is only an estimate, and can be tuned by the manufacturer to read just about however they see fit. For example it is common for consumer level GPS devices to base their accuracy reading on a 50% principle... If the GPS says the estimated accuracy is 20 feet, then there is a 50% chance the actual calculated position accuracy is 20 feet or better, but there is also a 50% chance the actual accuracy is worse (more) than 20 feet.
  • Boyd
    Posts: 11,613
    Barcard said:

    It is true that many addresses will be mapped a few houses off or even several hundred feet off.



    I'm pretty sure that individual addresses are not mapped at all, but that each road segment is allocated a block of addresses. The unit then estimates individual addresses based on your position within that segment.
  • Tim
    Posts: 19,394
    Boyd said:

    each road segment is allocated a block of addresses. The unit then estimates individual addresses based on your position within that segment.


    Basically we call this "interpolation" of address numbers. Not all streets are like that-- many addresses do get mapped with a greater precision than simple interpolation... Something I've found out while taking tours of Tele Atlas and talking with the NAVTEQ folks.
  • Boyd
    Posts: 11,613
    That is interesting Tim. I live in a very rural area, on a street with few houses. My driveway is a private right-of-way - #39a - which is located between #39 and #41. Neither my TomTom or Nuvi could find this address. The Nuvi was pretty accurate in finding #39 which is closest to my driveway but TomTom was off by quite a bit. When I moved here 3 years ago, none of the online mapping services could find 39A either.

    I just tried google maps now, and it now finds my address.... it doesn't flag the position of the driveway at the road, but it comes pretty close to the location of my actual house which is some 900 feet off in the woods! Obviously Teleatlas is doing their homework. 😃
  • Tim
    Posts: 19,394
    Neither company was willing to say about what percentage of street numbers are mapped with a greater precision than interpolation allows, leaving me to feel that the number was relatively low-- but both said that they are using methods more accurate than interpolation in some areas.
  • dsrussell
    Posts: 206
    I'm really looking forward to the article, Tim. I also find myself continually referencing your article about leaving your GPS on so it can download tracking data. It's such an important step and most people aren't aware that they need ... or should ... do this.

    I had guessed that the satellite data (accuracy within so many feet) wasn't without a liberal tolerance. When I see the number of feet, I also assume that is an estimated radius. Anything within 50 feet is pretty darn good in my book considering all the variables. It's only when accuracy gets to be a couple hundred feet or greater do I anticipate my Garmin thinking I may be on an adjacent road😃.
  • wscott52
    Posts: 393
    Tim said:

    Neither company was willing to say about what percentage of street numbers are mapped with a greater precision than interpolation allows, leaving me to feel that the number was relatively low-- but both said that they are using methods more accurate than interpolation in some areas.



    Good information here but it brings up a question I had from another thread. Is the interpolation of addresses done in the GPS internally or is it done by the map provider when they compile the map? I was under the impression the GPS unit just calculated a lat/lon fix and then does a look up in it's map database to know where to fix the position on the map. Is the GPS assigning addresses position using interpolation or looking up already interpolated positions?
  • gatorguy
    Posts: 7,244
    The map itself is the culprit. Your gps isn't ever looking for an address. It's only navigating to a set of coordinates associated with the map providers indication of where that address is.
  • alanb
    Posts: 2,233
    Since a map file is basically a data file and not an executable program, it seems to me that if interpolation (a calculated value from known reference points) is needed to find an address, the garmin software would need to do that calculation. The map data would of course be responsible for providing the nearest known reference points, but wouldn't garmin be responsible for the formula and calculation required to estimate the address coordinates?
  • Tim
    Posts: 19,394
    wscott52 said:

    Is the GPS assigning addresses position using interpolation or looking up already interpolated positions?


    I don't know for sure, but the database administrator in me says that the GPS is doing the interpolation. There would be quite a bit of data unnecessarily stored on the device if the map data contained the interpolated data. Might as well save tons of file space and let the CPU on board divide the number of addresses between two intersections by the distance between those intersections and apply the distance to each number. Again, I don't know that with certainty, but it just makes sense. The math is easy, the processor can handle the task, and there would be little discrepancy between devices on where they put each address even though it wasn't spelled out in the map database.
  • gatorguy
    Posts: 7,244
    That would be my guess as well Tim. I can't imagine the map provider having the time or inclination to set coordinates for every possible address on a minor street they never visited. Certainly much easier to give a known (??) start and end coordinate set on a street and let the device work out where the address might lie in between. Might be something to ask next time you're talking to one of 'em.
  • My 765T gets me to the house 2 doors down from mine when using the street address/house number, i can usually find my way from there👓
  • gatorguy
    Posts: 7,244
    A few times in my younger days that I was hard-pressed to find my way home from two doors away, so you're doing good. :mrgreen:
  • i still have issues in the fog :lol:
  • With all this in mind, I've always found navigating to addresses much more accurate than navigating to POIs. The POIs seem to have more error in their positions than the GPS' address interpolation.
  • We just got a 765t the other day and the accuracy in our 2003 built neighborhood is spot-on. The unit will put us right in the driveway, and sitting in the living room it will give our address when you go to the "where am I?" screen.

    What amazed me the most is the accuracy of the speed sign locations. Going from say a 50 to 35mph zone, the unit will change to the new speed limit on-screen within 20 feet after you pass the sign.
  • ordertwo
    Posts: 350
    gatorguy is 100% correct its NOT the GPS its the MAP DATA loaded on the GPS Plus One thing you have to remember too Is your Buying a $200-$400 GPS NOT a $50,000.00 Military Grade GPS So its not going to be 100% accurate all the time .
  • greentube
    Posts: 2
    wrong thread
  • lordgrinz
    Posts: 132
    Maybe the answer is associating Lon/Lat to all physical address info?
  • gatorguy
    Posts: 7,244
    If it's still interpolating from a range, theu same issues would exist. Physically visiting each and every residential and commercial building and inputting those coordinates is not going to happen
  • lordgrinz
    Posts: 132
    They already do this, its called the US census, just a matter of giving them proper tools to enter the info.

    gatorguy said:

    If it's still interpolating from a range, theu same issues would exist. Physically visiting each and every residential and commercial building and inputting those coordinates is not going to happen

  • gatorguy
    Posts: 7,244
    Hmmm. . . That actually would be a good suggestion.😀

    I remember reading that the census canvassers have an enhanced software they should be using this go-round. Wonder if that is being logged for government use.

    Now actually including that amount of data within the mapset might push the size beyond what we can currently use, but with a detail trade-off (even higher compression) maybe could work. I don't have any idea how much something like that adds to data size.
  • My Garmin 760 gets me pretty close, so no complaints. Once I do get to where I am going, if I know I will be going there again, I will use the "Where AM I" feature and save the position from there. Our daughter is in competitive cheer and we have a pretty good sized carpool. With this, I will save it from "Where Am I', then save it to Favorites. That way, when my wife drives, she doesn't have to worry as the GPS will get her right to the front door.
  • wfooshee
    Posts: 114
    I know for a fact that my device (Streetpilot 2610) interpolates addresses, as my house is the corner lot at the highest number on the block. 4035, my actual street number, shows about a 3rd of the way down the block. 4099 or 4095, somewhere like that, gets in front of my lot. When someone is coming in from out of town, I give them my correct address, then tell them to put 4099 into their GPS if they use one.

    Google Maps used to do that, but now has plots divided in my neighborhood (and most of my city) so it goes right to my address. I think their servers might have more storage space than my 2610, though. 😃

    So each stretch of road in the map knows its address range, but not the actual addresses. That would inflate the database to about 30 or 40 times as many "address" records as it has now, needing to flag smaller and smaller segments of streets rather than just segments between intersections.

    Interpolation might be made more accurate without recording each and every plot if maybe the map contained the max address of each segment, rather than assuming XX99.
  • I just read through this thread for the first time and may be able to clarify/confirm a couple things.
    First I am a GIS cooordinator and work almost exclusively with street centreline files...the same type that are used in GPS units, although I am in gov't so not affiliated with Navtek or other commercial mapping companies.
    The street files most commonly use a block range for addressing. So each street segment (a segment being defined as one piece of street between two intersections) carries 4 addresses. A 'from' address and a 'to' address for each side of the street. So, to determine all addresses along that segment interpolation usually occurs. For instance, lets say your street segment has an address range from 100 to 200 (on one side) and 101 to 199 on the other side...odds on one side and evens on the other. if you search for 150 the GPS will put you halfway down the street. That may, in realiaty not be exactly where 150 is, but most algorithms search that way. Its really simple math. In reality it could be that all of the houses are crammed at one end of the street and if thats the case the error in interpolation will be greater.
    A very good reason ranges are used are that it is that it is much easier to keep up-to-date. If individual points for each address were used it would be a nightmare to update every time a new house was built along a street.
    Hope that helps a bit.
  • Boyd
    Posts: 11,613
    Very cool - thanks for your insights.😃

    Are you involved with updating the TIGER files perhaps? I make my own maps, and it sometimes surprises me how many short line segments are used within a street.
  • That makes sense, always wondered why all the mapping software, including google, put us around the corner from our house...
    Guess that definition of a segment adds another variable to it if it is curved roads or circles...

    always thought they used from one corner to the other corner as the min and max for a segment, but in our case our street is named a circle that is more like a regular block, i.e. a U... we are near the mid point of the bottom of the U...

    given that, I assume the min and max is from the 'top' of either side of the U and it guesstimates based on house addresses how far around the circle we are....

    How about that - just looked @ google for the first time in a while and it now puts our arrow away from the street, very close to the middle of our back yard😃 but at least in the correct address now !
  • Boyd said:

    Very cool - thanks for your insights.😃

    Are you involved with updating the TIGER files perhaps? I make my own maps, and it sometimes surprises me how many short line segments are used within a street.



    No, I'm in Canada.
    As for your observation on short line segments, there could be a number of reasons for short line segments. (By shorter line segments I assume you mean a segment that is broken between intersections.)
    It depends entirely on the data set you are looking at and how it was designed, but they could be broken more often for a nmber of reasons. Possibly a change in road surface, a municipal (or some other boundary) line, change in speed of the road, etc, or often for no apparent reason at all.
  • always thought they used from one corner to the other corner as the min and max for a segment, but in our case our street is named a circle that is more like a regular block, i.e. a U... we are near the mid point of the bottom of the U...

    given that, I assume the min and max is from the 'top' of either side of the U and it guesstimates based on house addresses how far around the circle we are....


    Each line segment has a direction, a start and finish point, and that determines the direction of that line segment. (this has nothing to do with one way streets, etc). Usually the direction of the segment corresponds to the the low to high addresses for that street, although it doesn't necessaily have to. Its just easier to work with that way.
    So the low odd and even address for that street are at the start point and the high odd and even address for the street are at the end point of that segment.
    Given that, where your house is placed along the street depends entirely on the low and high address for that segment and the length of the segment. As I said before, very simple math, which I assume the GPS processor performs on the fly. It shouldn't never vary unless the map changes.
    Don't confuse this with where it places YOU when you are looking at the live GPS point...meaning your car. THAT can vary depending on how accurate the GPS receiver is feeling at any particular time. And don't ask me about GPS accuracy, I deal mostly with the mapping end, not GPS technology.
  • wfooshee
    Posts: 114
    Google is putting actual lots on their maps, they started doing so some time back, but their database can be very much larger than anything you can carry around in your GPS receiver, and they've got a pretty good indexing system they've developed for searches, as well. Seeing as how indexing and searching is where they come from, right?

    Anyway, going to my address in Google Maps goes to my lot, a corner lot at the high end of the address range for my block, 4035. Any other search system I have to put 4099 to get to the end of the street.

    Basically, most mapping databases we have these days don't have actual physical addresses mapped, only the range on any given block. This is the 1500 block, so it has 1500-1598 on one side, and 1501-1599 on the other. Even if the block only had 7 houses on it, you'd have to be at 1580-something to get to the high end of the block.
  • wfooshee said:

    Google is putting actual lots on their maps, they started doing so some time back, but their database can be very much larger than anything you can carry around in your GPS receiver, and they've got a pretty good indexing system they've developed for searches, as well. Seeing as how indexing and searching is where they come from, right?


    That doesn't surprise me at all, google is doing some pretty cool stuff with thier street views, etc. And they seem to have the resources to do it.
    But what they are doing is far different than what you see in a Garmin or Tom Tom, the data and technology.
  • r.segall
    Posts: 1
    I hvve alsp experienced this problem with my nuvi, Garmin has notyet corrected it,acting on my complaint.
  • yorkierver
    Posts: 1
    I am a full time RVer and use my Garmin almost daily. It has given me directions to turn the wrong way on one way streets, sent me to a cemetery when looking for a laundromat, directs me to turn right out of a parking lot going 2 blocks or more in the wrong direction then directing me to loop around a block and go in the opposition direction rather than just turn left out of parking lot. This can often put me in a precarious situation when pulling an RV. I had a Magellan and rarely had this issue. This seems to be a real problem with Garmin. It will place me on a route and then suddenly recalculate my directions. It has instructed me to exit an Interstate when I know that is incorrect. I do not feel confident following the directions with this Garmin. I am currently in Georgia and have found numerous map errors in this state. Any recommendations or possible updates to correct these issues?
↑ Top

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!