12
$\begingroup$

Obviously GPS is the most obvious and accessible technology for obtaining a locational "fix" for a robot at any particular time. However, while it's great sometimes, in other locations and situations it's not as accurate as I'd like, so I'm investigating whether there's a relatively easy way to improve on this accuracy (or not, if that turns out to be the case.)

I've considered the following options, but found limited information online:

  • Would using a much better antenna help, especially for low signal areas? I'm thinking yes to this, but if so how would I construct such an antenna and know that it's an improvement? Are there any good guides on how to do this? I could use a ready made antenna if they're not too expensive.

  • Would using multiple separate receivers in tandem help, or would they likely all be off by a similar amount, or would I not be able to extract a meaningful average with this approach?

  • What sort of characteristics should I look for when choosing a good GPS receiver to help accuracy?

  • Is there anything else I should consider which I've missed?

Improve this question
$\endgroup$
1
  • $\begingroup$ You certainly should try to get a dual-frequency receiver. Apart from that, what exactly is your goal with the locational fix? If, say, you want to get back to the place you started from, some camera and image analysis may get you from meters down to millimeters of accuracy much easier than anything satellite-based. $\endgroup$
    – Christopher Creutzig
    Commented Feb 23, 2014 at 14:24

5 Answers 5

Reset to default
8
$\begingroup$

Can't answer all your questions, but based on your use case Differential-GPS might help you. Modern tractors are using this for precisely navigating on fields (in autonomous mode). Here fixed ground stations are used, which know their exact position and calculate the error in the current signal. This adjustment is then used by the other GPS receiver in the area.

Improve this answer
$\endgroup$
1
  • $\begingroup$ D-GPS would indeed vastly improve your accuracy, by about two orders of magnitude. $\endgroup$
    – ThomasH
    Commented Oct 24, 2012 at 10:19
6
$\begingroup$

A traditional approach is to use an error correcting algorithm like a Kalman filter. By combining dead reckoning from wheel encoders and heading commands with GPS you can smooth out GPS jitter. This is not an instantaneous improvement as it requires a series of measurements to estimate the error inherent in the GPS signal.

Improve this answer
$\endgroup$
1
  • $\begingroup$ Wow, Kalman filters, that takes me back a few years. *8') $\endgroup$
    – Mark Booth
    Commented Oct 23, 2012 at 23:39
3
$\begingroup$

As other posters have pointed out, using some form of differential GPS will give you large improvements in precision. We have used the Magelan mb500 commercial platform and it claims an accuracy of around 2-3 cm when in RTK fix mode. However, in areas of bad satellite coverage it would sometimes not even provide a solution at all, where cheap GPS receivers in a phone could.

There is an open source library called RTKLib which looks very promising for using cheap GPS equipment with a number of different methods for improving the precision.

Improve this answer
$\endgroup$
1
  • 1
    $\begingroup$ For our field rover we use the Ashtech MB100 link It requires a basestation for RTK fix (also an MB100 board) this basestation must know its real location, and must not move. The basestation sends correction signals to the rover over a wireless link (it's serial data, you choose the link). This is a rather expensive solution, but we have 1-2 cm accuracy relative to the basestation, absolute precision depends on the precision of the basestation placement. $\endgroup$
    – Karl Damkjær Hansen
    Commented Oct 24, 2012 at 12:22
3
$\begingroup$

I actually work as a programmer for the tractors that bit-pirate mentioned.

There are several ways mentioned that can get you more accurate results. It depends largely on your application and what you are trying to accomplish.

  1. WASS uses a signal to augment not as accurate as some other methods but you will get decent.

  2. If your platform is fixed you can just average the position over time and get good results.

  3. RTK is very accurate, and the library mentioned by Jakob looks pretty cool. Some states like Iowa have a government run RTK system that you can use.

  4. I found this the other day for dead reckoning.

  5. On a mobile system the antennae offset to the center of gravity will effect the accuracy. E.G. On a steep hill the GPS antennae will be off center.

What are you trying to accomplish? The atmospheric distortion for multiple antennas in the same location would be very similar.

Improve this answer
$\endgroup$
1
$\begingroup$

If you want to be really clever, you could consider Assisted GPS (see Wikipedia for full details) which uses the phone network to provide a better initial fix, by using the cell towers fixed positions as a basis.

Note, however, that A-GPS uses your data stream, so if you have to pay for data, you may not want to use this!!!

Improve this answer
$\endgroup$
1
  • 1
    $\begingroup$ Assisted GPS can't really improve the positioning accuracy which is what I'm after, it really just improves the startup time to get an initial fix. $\endgroup$
    – berry120
    Commented Oct 24, 2012 at 10:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.