Shop OBEX P1 Docs P2 Docs Learn Events
Parallax GPS Receiver Module Accuracy — Parallax Forums

Parallax GPS Receiver Module Accuracy

edited 2007-07-05 10:49 in Robotics
I was just curious as to how accurate the GPS Reciever is (in inches/feet).· Is it possible to construct a series of way points for my bot to follow?· Would like to know without spending the $$$.

Comments

  • CCraigCCraig Posts: 163
    edited 2007-06-28 16:37
    It's on page four of the product specs.

    See this thread.

    http://forums.parallax.com/showthread.php?p=657666
  • allanlane5allanlane5 Posts: 3,815
    edited 2007-06-28 20:01
    Sheesh. +- 5 meters.
  • fishfish Posts: 55
    edited 2007-06-29 02:47
    +- 5 x (3.24077649
  • FranklinFranklin Posts: 4,747
    edited 2007-06-29 03:21
    Yes and it's better than Garmin or Magellin units.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    - Stephen
  • OSOKOSOK Posts: 35
    edited 2007-06-29 05:16
    Um, the foretrex 201, a $100 dollar GPS by Garmin has an accuracy of within 2 meters in horizontal position and 4 meters in vertical.
  • LSBLSB Posts: 175
    edited 2007-06-29 12:48
    If you just build a REALLY big robot...
  • BeanBean Posts: 8,129
    edited 2007-06-29 14:22
    OSOK said...
    Um, the foretrex 201, a $100 dollar GPS by Garmin has an accuracy of within 2 meters in horizontal position and 4 meters in vertical.
    I downloaded the manual for the Foretrex 201, on page 54 it states the position accuracy as 15 meters (49 feet).

    Bean.


    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    “The United States is a nation of laws -· poorly written and randomly enforced.” - Frank Zappa

    - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
    www.hittconsulting.com
    ·
  • CCraigCCraig Posts: 163
    edited 2007-06-29 15:48
    I've used the Parallax unit side by side with a Garmin 60C. When marking a trail they both are marking the same trail. So much so, that the only one I use for logging now is the Parallax one. It is in the Jeep, mounted, and records 24/7. When in a forest, or worst when I'm in a jungle, the Parallax unit is the last to drop signal and the first to get it back. The Sony radio they use (can't remember model, but saw a message from it when I was re-programming it) seems to be a good radio. I know that there are some really hot new radio chip coming out, but I don't think that even the big guys use them.

    I'm very happy with the unit. That and the fact that you can re-program it as you see fit. I might have to get another one.

    HTH, Chris
  • LarryLarry Posts: 212
    edited 2007-06-29 17:15
    So--

    How often can you get readings from the Parallax unit? can the list sentences it outputs be edited?

    I've been using a Garmin unit, but it outputs so many sentences, I can only get the GPRC once per second.
    I'd like to do better.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
  • Chris SavageChris Savage Parallax Engineering Posts: 14,406
    edited 2007-06-29 17:36
    Larry,

    I have never seen a unit that refreshes more than once per second. That’s not to say they don’t exist, but at 4800 bps sending out the data and all the $GPxxx strings there just isn’t enough time to send it more than once per second. To answer your question though, the Parallax unit parses the $GPRMC strings and provides the data requested. It’s not possible to get newer data more than once per second. Take care.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Chris Savage
    Parallax Tech Support
  • LarryLarry Posts: 212
    edited 2007-06-29 21:48
    Thanks Chris,

    There are units where you can restrict them to one or two sentences, and they'll just send them over and over. Trouble is, the only ones I have seen are Sub-meter resolution units, and they cost $2400 new, and ocasionally about a grand on Ebay. With one sentence, they'll work at about 10Hz.

    There are alot of units out there, however. I keep hoping someday someone will build a cheap OEM unit that will allow that kind of customizing.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


    Post Edited (Larry) : 6/30/2007 3:40:00 AM GMT
  • BeanBean Posts: 8,129
    edited 2007-06-30 01:24
    Larry,
    · Here is a unit for $100 that does 5Hz update rate (5 readings per second).
    · http://www.sparkfun.com/commerce/product_info.php?products_id=8266

    Bean.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    “The United States is a nation of laws -· poorly written and randomly enforced.” - Frank Zappa

    - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
    www.hittconsulting.com
    ·
  • LarryLarry Posts: 212
    edited 2007-06-30 03:49
    Thanks, Bean!

    All the sentences at 5Hz, though you'll need to read at 38kB. It might have to go on a Prop. They claim indoor reception!!! I don't need that, but it might help in the trees.
    3.3m error circle. Not sub-meter, but not bad.

    Looks like a nice unit do you have one?

    It is configurable as to which sentences it outputs. I'm only interested in GPRMC and GPVTG sentences.
    I wonder if I shut off the rest if I can get faster than 5Hz?
    Even if that's the peak, it might be good enough, if I slow down my robot a bit.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
  • lboucherlboucher Posts: 139
    edited 2007-07-04 01:16
    Hey All

    I work in the robot biz, and even 50,000 dollar units have on the range of 5 meter accuracy. Its just the nature of the beast, Inertial Navigation can get that accuracy down to the meter range typically, but can still be higher in certain situations.
  • BeanBean Posts: 8,129
    edited 2007-07-04 01:45
    Larry,
    No I don't have one so I don't know about your questions. I just remembered that they had one that was better than 1Hz update rate.
    Let us know if you get one, and how it works for ya.

    Bean.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    “The United States is a nation of laws -· poorly written and randomly enforced.” - Frank Zappa

    - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
    www.hittconsulting.com
    ·
  • LarryLarry Posts: 212
    edited 2007-07-04 04:40
    iboucher said...
    I work in the robot biz, and even 50,000 dollar units have on the range of 5 meter accuracy. Its just the nature of the beast, Inertial Navigation can get that accuracy down to the meter range typically, but can still be higher in certain situations.


    Somebody better tell these guys----www.starpal.com/Reports/Submeter_GPS_Report_e.html 1.89 meters with no DGPS,WAAS, Beacons, or Omnistar. And it's not the best.If you hunt around, there are some that claim better repeatability than this. It lists for about $3000, cheaper on Ebay.

    What I'd really like is one of those silver hockey pucks from Tomahawk missiles that the Whitaker teams flash around.
    Can I get one for under $200?

    Repeatability is what I seek.


    I'm still looking for something I can afford. It might be the sparkfun unit.

    I looked again at the specs for the Parallax model- specifically at the "smart mode". Looks like a nice feature.
    If I'm just querying one of the parameters, can I do it at faster than one Hz? If so, Is Baud rate the only limiting factor? Anybody have one of those? The datasheet wasn't very specific.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


    Post Edited (Larry) : 7/4/2007 4:45:15 AM GMT
  • CCraigCCraig Posts: 163
    edited 2007-07-04 12:42
    Larry,

    The way the control program is written, you would only get data out of the Parallax unit once a second. The program, in smart mode, waits for your command, then waits for the radio to send current data (1 Hz), then sends it to you.

    You could re-write the GPS control program to do what you want. However, sub second requests would give you 'buffered' data. The radio only sends new data to the control program once a second.

    HTH, Chris
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2007-07-04 16:37
    One possible consideration with the Parallax unit's smart mode that the datasheet doesn't address is the following: Since there's no command to "freeze" a reading, it seems likely that consecutive latitude and longitude requests, say, might occasionally retrieve data from different NMEA sentences. If this is the case, you'd have latitude from one position and longitude from another. The way to overcome this would be to take both readings twice in rapid sequence and compare the two. If they're not identical, do it again until they are.

    -Phil
  • LarryLarry Posts: 212
    edited 2007-07-04 20:46
    CCraig said...

    The way the control program is written, you would only get data out of the Parallax unit once a second. The program, in smart mode, waits for your command, then waits for the radio to send current data (1 Hz), then sends it to you.

    Thanks for responding (and PhiPi), but It just got fuzzier. Does that mean that is you ask for Longitude, then Latitude, (note that this is the reverse order from what a $GPRMC sentence gives out) It takes 2 seconds? That doesn't sound right (If it's true, then Smart Mode isn't so smart). I'm hoping you mean it reads from values stored in a buffer, even if it's only getting updated once per second.

    If someone could ask for a time stamp in smart mode, then Longitude, then Latitude, then another time stamp, it would clear that up. Knowing whose GPS and chipset Parallax is using would also help, allowing a peek at the datasheet. It might well be possible to get readings faster than once per second with a little hacking.

    That would also help with how to deal with PhiPi's issue about not reading from the same sentence. If it is reading from a buffer, you could just ask for the time twice, as above, to see if it has been updated.

    It actually may not be such a big deal for my purposes. If you are navigating a moving robot, you probably want the latest available reading anyway, and if you're still, it shouldn't matter much .

    But getting a GPS that actually reads data more than once per second is really what I'm after. One of the reasons I'm so interested in the read rate is that I want to try reducing the error circle by doing some Least squares averaging of multiple readings once I get near a waypoint. I'd like to do that in a second or two, not take half a minute. There are reasons to want to be able to do it while you are rolling, too.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2007-07-04 21:27
    Larry,
    Larry said...
    Does that mean that is you ask for Longitude, then Latitude, (note that this is the reverse order from what a $GPRMC sentence gives out) It takes 2 seconds?

    Not at all. I assume all readings are buffered until the next reading is gotten from the receiver. But if that reading should come in between requests from your program, one request will get old data and one will get new data. Since latitude and longitude are retrieved by separate requests, the possibility of their being split by new data exists. However, the strategy that I outlined above should eliminate any effects of this.

    Caveat: The documentation is extremely sketchy on these details, so I'm reading between the lines here.

    -Phil

    Post Edited (Phil Pilgrim (PhiPi)) : 7/4/2007 10:33:51 PM GMT
  • LarryLarry Posts: 212
    edited 2007-07-04 22:41
    PhiPi said...
    Caveat: The documentation is extrememly sketchy on these details, so I'm reading between the lines here.

    Yeah, I hear you , Phil, but your post does clear things up a bit.

    I guess the folks trying to get the product out the door can't anticipate all the ways we want to use the products. Sometimes you just have to forge ahead with the info you have.

    If the Smart Mode does what I think it does, it makes the product very appealing, reducing by at least one the number of processors I've had to put on a mobile robot.
    fixed byte parsing and quick reads sound pretty nice. It'd be nicer if I could figure out how fast it was.

    In the past, I've had to devote one processor excusively to Parsing sentences and extracting information,passing it on, then waiting for the next read cycle.


    What to do----What to do----

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2007-07-04 22:49
    Larry said...
    It'd be nicer if I could figure out how fast it was.
    Chris Savage (Parallax) said...
    It’s not possible to get newer data more than once per second.
  • LarryLarry Posts: 212
    edited 2007-07-04 23:00
    Yeah, but how many times a second can you read the same data? That determines the rate of your steering cycle.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2007-07-04 23:40
    That would be a function of the baudrate (4800) and the number of bytes exchanged during each request. Since that latter number is typically very small (6 each for latitude and longitude), multiple reads per one-second "epoch" seem highly likely.

    But if your program reads data that often, it will see discrete jumps in the current position and will have to apply filtering to it anyway to obtain a smooth steering control. And bear in mind that even when the position does change, the new position won't be your real current position but the position computed by the receiver maybe a half second ago after its latest satellite fix and before it output the NMEA sentences to the onboard micro. So regardless of the read rate, there will always be a lag to account for in your control algorithm.
    Larry said...
    What to do----What to do----
    'Sounds like you just need to try it! At only $69.95, how can you lose? wink.gif

    -Phil
  • phil kennyphil kenny Posts: 233
    edited 2007-07-04 23:48
    Here is a program that runs on a PC and plots the output from
    a GPS receiver:

    http://www.huntting.com/sawatch/

    With the receiver stationary, looking at the output over a 24 hour
    period can give you a pretty good idea of the repeatability you
    can expect.

    Note that I said repeatability, and not accuracy. To determine
    accuracy you need to know the exact longitude and latitude
    of your location. Performing this test at a USGS Geodetic
    Survey marker is one way to determine the accuracy of your
    receiver.

    Here is a link to locate the USGS marker nearest you:

    http://www.ngs.noaa.gov/cgi-bin/ds_radius.prl

    phil
  • LarryLarry Posts: 212
    edited 2007-07-05 05:03
    Phipi said...
    But if your program reads data that often, it will see discrete jumps in the current position and will have to apply filtering to it anyway to obtain a smooth steering control. And bear in mind that even when the position does change, the new position won't be your real current position but the position computed by the receiver maybe a half second ago after its latest satellite fix and before it output the NMEA sentences to the onboard micro. So regardless of the read rate, there will always be a lag to account for in your control algorithm.

    All true, but one of the beauties of GPS is that you can be off a bit, sometimes radically so, and the GPS will inexorably draw you into the error circle. I once had a compass that was reading (mostly randomly) 90º off from where it should have been, and still won the contest. The robot still entered the error circle. it just did so in a big arc insted of in a line. That scheme works best if you have a bunch of chances to correct, even if they are a bit wrong.

    As to filtering, you have to do that anyway. Most guys, including me, steer by compass, GPS heading being inferior at slow speeds. One of the features of most contest venues is that they are in urban areas subject to underground cables, steel and other disruptive influences. You need to filter the steering over at least a couple seconds in order to pass over these obstructions. Ignore that issue, and your robot can head out of sight down the cable.



    I could be wrong. Remember that this day in 1776 George III entered into his diary that "nothing of importance happened today"

    fireworks are starting.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
  • LarryLarry Posts: 212
    edited 2007-07-05 10:49
    phil Kenny,

    Thanks for the links. I'll have to give SAWatch a try. Repeatability is OK for a RoboMagellan contest. You are allowed to take your own readings on the targets before the contest.

    I've found Google Earth is a very accurate means of calibrating your GPS as well. I've checked the coordinates of a geo marker in a park nearby and the numbers were spot on. Same with a city survey pin on my corner. I got the coordinates from a surveyor one day.

    Distances can be measured very accurately, also. The first time I measured a baseball diamond wth GE, the numbers came out short, and the distance between first and home was different than between first and second. Turns out they are supposed to be, by just what I measured with GE. The 90' square is measured to the outfield sides of first and third, but to the center of second base. The foul lines on "my" field are exactly NSEW, too.

    If you have access to a football field, they'll work, too. Ready made calibration grids that often go EW.


    Anyway, I'll probably get the Parallax GPS. If it doesn't work out for the robot, I have another use for it.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Sign In or Register to comment.