Flutterby™! : GPS questions 2009-02-23 17:30:43.260078+00

GPS questions

2009-02-23 17:30:43.260078+00 by Dan Lyke 13 comments

This is kind of a two-part question, and probably both of them are for Eric.

I have an ancient Garmin hand-held GPS. I can download, via the serial port, in almost any format I want (yay for gpsbabel), but waypoints and trackpoints end up at different parts of the file.

Eric and I figured out how to make OpenLayers and Google Earth work to let me draw out hiking and biking paths and, with a little yet-to-be-written Perl drop them into some web pages I'm playing with, but I think I also need a tool that coalesces those waypoints and trackpoints by time into a single stream, and crop to them. I don't want the "driving to the trailhead" portion.

And the second question is: Is there a later GPS device that'd make this simpler? Something that just stores tracking info, maybe has a button that lets me mark trackpoints, that charges via USB and I can plug into my server (or even my Windows box) and set up a script to download the stuff and upload it appropriately?

So more seriously, it looks like Google Earth (almost typed Gogol Earth... geography of a madman, anyone?) will do some editing. Not quite what I wanted, but a start.

For trimming your routes, try OpenJump. It does the best job of simple vector data editing of any GIS I've used.

I don't use very many different consumer GPSes - but they all seem to have a similar interface (all equally convoluted) for starting and stopping tracks. My recommendation: import your GPS tracks into OpenJump and clean them up. Or load your GPS tracks and use them as a visual guide to heads-up digitize (i.e., draw) your route where you need it.

Umm, Perhaps I'm simpleminded, but I use "LoadMyTracks" to pull the data from the GPS. It gives me a GPX or KML file. The latter, being more or less XML, allows me to edit with a text-editor. A guy could probably script it with [your favorite language here]. Then feed the file into Google Earth, and out squirts far more information than I want.

What I want is a replacement for that last step. I want an easier way to turn the file of points into a jpg with my map on it. I don't want to zoom around or navigate on the fly. I just want a map that shows where I went. If I had some control over the scaling and such, that'd be cool, but if it just made web-ready graphics, I could work with that.

Dave, the issue I've had with editing the text file directly is that it takes a bit of cognitive load to correlate the trackpoints and the waypoints and I just didn't want to bother with writing the "what went where" bits.

On the points in a JPG with the map, OpenLayers can give you the tiled one in a web browser. Somewhere around here I've got code to grab images from TerraServer and plot points on 'em, it shouldn't be hard to save that back out.

Eric pointed out, and I didn't understand just how bad it was 'til I was playing yesterday, that some of the image to vector alignment in the Google data set is atrocious.

This article reminded me that I wanted to look into a high-accuracy GPS device to help me work on my property. So after Googling around, with the hairs on the back of my neck raising from sticker-shock, I discovered the least expensive option with the sort of accuracy I'd need was $2 grand.

Yikes. Still, it'd cost about that much to get a surveyor out for a single visit. But why should this equipment be so expensive?

Shadow, I believe because that kind of accuracy either requires something that'll sit for a few weeks, or needs a separate terrestrial transponder that you can put in place and let sit for few weeks.

GPS as we consumers deal with it sucks, it's just that there are enough heuristics we can apply to it and the data we're using it to reference it that it's usable for some stuff.

And when you're paying for the surveyor, you're also paying for their liability insurance and credibility in court.

Dan, they've stripped most of the obfuscation from the GPS signal, and with the right hardware you can get half-decent accuracy from a single portable unit.

This is the unit I've been looking at; it's rated by the manufacturer as giving real-time accuracy of better than 1 meter resolution. It's not survey-accurate, but it is good enough that I can use it to plan what I need to. Part of the expense, I'm sure, is that it has tons of bells and whistles, like displaying aerial imagery, when all I'd really need is a coordinate readout...

Shadow, it's been a few years since I've ducked into this heavily, but at the time I was looking at military applications, and, yeah, all of my numbers were with selective availability turned off (It's been off since the beginning of May 2000). There may have been some advances in inertial information and wavelength phase detection since then, and I know there are a bunch of terrestrial transmitters now that are primarily used for aviation applications (because altitude in satellite GPS can easily have 30 meters of noise), but as of 4 or 5 years ago or so I was told that 95% of samples would be within 10m, so there were a lot of heuristics based on what was known about the geometry of the terrain to, for instance, try to figure out which side of a wall the device was on.

Here's a good discussion of error accumulation in satellite GPS, you can hit the table at the end that has a list of sources of inaccuracies that sum to ±15m, but I'll bet distribution of those errors isn't a straight summing. If anyone's getting better than that (and I've talked to guys working on agricultural tractor guidance systems who are measuring their error in centimeters with the help of additional terrestrial signals, so people are) they're doing something other than just watching satellites.

Their real-world test gives an accuracy better than 2 meters, with varying results from depending on different beacons. (According to the Coast Guard there aren't any public beacons covering me; sucks.)