Accelerometer for track smoothing?

[+user13371]

Are there any handheld GPSRs that include an accelerometer to aid in track smoothing?

 

I know DeLorme PN40 and Apple iPhone include accelerometers, but I don't know if they're used to discard or correct bogus track points.

[+embra]

AFAIK, Lee, the PN-40 doesn't use the accelerometer for anything beyond stabilizing the magnetic compass readout. Not having been privy to the design meetings, I could be wrong about that.

[+EScout]

I thought this was a mechanical device, but I think firmware can probably work in a GPS since it measures location and time. I assume you are asking because you want a better track than you are getting.

 

One of the best features of the Magellans such as eXplorists, is the track log. I use mine several times a week, and it leaves a nice track with no spikes or bogus detours. I can stop and take photos and the track stays put until I move again. Great for geotagging your photos and putting tracks on Google Earth in KML files.

Edited October 14, 2008 by EScout

[+user13371]

I assume you are asking because you want a better track than you are getting

Good assumption. I actually find my Legend HCX to give a better track log than my eXplorist 210, but neither are completely free of the occasional "spikes" you mention.

 

In another context, I've seen apps on the iPhone that really show off how sensitive a cheap accelerometer can be (e.g. a siesmometer showing the vibrations of someone walking by).

 

The "spikes' you see in track logs are just calculated fixes that are wrong. By adding in data from an accelerometer, a GPS could br programmed to smooth those out or ignore them in the track log.

 

Was just wondering if there are any handheld, consumer grade devices that already do this.

[+trainlove]

Anyone ever try to look at the vertical profile of a track on a Magellan that does not use an altimeter. I once took a nice, leasurly 10 hour kayak down 12 miles of river, not 10 miles from the ocean, and the track log looks like hell, +/- hundreds of feet spikes all over the place. I think I posted that in my gallery photos for that event.

 

Now if there were such a thing as a GPS with a real accelerometer in it, then one would have a Inertial Navigation System and perhaps not be limited by the precision of the GPS system.

[Grasscatcher]

I assume you are asking because you want a better track than you are getting

Good assumption. I actually find my Legend HCX to give a better track log than my eXplorist 210, but neither are completely free of the occasional "spikes" you mention.

 

In another context, I've seen apps on the iPhone that really show off how sensitive a cheap accelerometer can be (e.g. a siesmometer showing the vibrations of someone walking by).

 

The "spikes' you see in track logs are just calculated fixes that are wrong. By adding in data from an accelerometer, a GPS could br programmed to smooth those out or ignore them in the track log.

 

Was just wondering if there are any handheld, consumer grade devices that already do this.

 

Lee,

Here's another unsolicited opinion...

I recently was trying to edit /smooth some tracks that came from a legendHCX, and was amazed to personally see just how many random and unexplained errors there were. They absolutely were not due to operator error. I had heard of the "drift" problem with all E-Trex H series but had no idea just how "sorry" the tracking results were, as compared to tracks recorded with my 76CSx. I feel that the problem chipset for the H series is totally unacceptable for recording accurate tracks right now. To be corrected in the future? Who knows!

Have you personally compared your tracks to some that were recorded on a 60 or 76 x series? Do you still see the problems?

 

There is a software program by the title of Expert GPS by Topografix, that is the absolute best (my opinion)

for editing tracks.....even down to working with individual trackpoints. A user that is familiar with that program can quickly edit a track where it appears as smooth as the proverbial Baby's Behind.

[+user13371]

...another unsolicited opinion...

You said it - not me

 

I do appreciate the comments on how good or bad the tracks are on a given model of GPS, and what the best software might be to adjust/fix them, but that really doesn't speak to the original question. I'd just like to stay on topic.

 

A GPS with even a cheap accelerometer - like you find in an iPhone or a WII controller -- could be smart enough to know when it's standing still (or at least moving quite smoothly) or being jerked around. And that could go a long way to dropping erroneous points from a track log.

 

I'm surprised the Delorme PN40 (which supposedly has the hardware to detemine its orientation) doesn't also use it for that purpose.

Edited October 14, 2008 by lee_rimar

[+LEGO]

I think these new Sony GPS'ers have what you are looking for.

Here is a direct link to the Sony site.

[+user13371]

I think these new Sony GPS'ers have what you are looking for...

Hmm, ad copy does hint at it. Hard to tell for certaind as they're not shipping products yet. And they look more like vehicle nav systems than hand-helds....

[+PDOP's]

Your accelerometer idea is interesting. Another approach might be some sort of speed filtering in the firmware of the GPS. I just reviewed a recent tracklog of a walk and as is typical for my Garmin76C the inaccurate segments of the track are long zig-zags which have indicated speeds of 80 to 90 KPH.

 

Being able to set a speed filter which would ignore segments with speeds over a set value would go a long way towards creating better tracklogs.

[+GOT GPS?]

My most accurate GPS to date, seems to be the Map60CSx, and when held as vertical as possible, I got excellent tracklogs. My Map60Cx without the sensors, gave a pretty terrible tracklog, where would be all over the place with the elevation, even when walking on flat land.

 

My Explorist XL had problems at times picking up bad signals, and using them. One time as I was hiking with my friends, on an island in Put-In-Bay Ohio, the GPS jumped like 12 miles back to land, then back to reality, in a matter of seconds.

 

Hopefully, the manufactures can include a speed filter on the Satellite Page int he Future, that you can set an upper limit, that you know the GPS will not speed past. Like selecting 5mph max, then the GPS rejects all readings that indicate a movement faster than 5mph. I would also put into GPS software, an accuracy filter, to reject readings worse than a user selectable amount. Like if you selected to reject readings worse than 20 foot accuracy, then the GPS will just sample the signals again, till the readings get below 20 foot accuracy. This way there would be no spikes in the tracklog. The worst spikes in a tracklog where from a Colorado 400t, and after walking down my sidewalk. The Colorado 400t showed a severe sawtooth tracklog, whereas the sidewalk was perfectly straight.

 

By the way that Seismometer is now on my iPhone 3G.

[+Red90]

Magellans used to do those real time filtering. That made them horrible for caching, IME...... A sensitive and quickly changing tracklog is very helpful to cache searching....

[+trainlove]

Magellans used to do those real time filtering. That made them horrible for caching, IME...... A sensitive and quickly changing tracklog is very helpful to cache searching....

There IS a difference between tracklogs and position update rate, apples vs oranges.

[+Red90]

Magellans used to do those real time filtering. That made them horrible for caching, IME...... A sensitive and quickly changing tracklog is very helpful to cache searching....

There IS a difference between tracklogs and position update rate, apples vs oranges.

 

Yes, I know.... I did not start caching yesterday. Looking at the tracklog gives a good understanding of the search area you have covered and if the average position fix is actually where you are searching.

[+Klemmer]

If you want to read about the leading edge (bleeding edge?) technology of GPS, Inside GNSS is really good. I happened to remember an article about inertial sensors being used with GPS receivers: Integrating Inertial Sensors with GNSS Receivers. Enjoy.

[John E Cache]

Are there any handheld GPSRs that include an accelerometer to aid in track smoothing?

 

I know DeLorme PN40 and Apple iPhone include accelerometers, but I don't know if they're used to discard or correct bogus track points.

Interesting idea. How does an accelerometer help if you can point the GPS any direction? Don't all of the points have errors and could be considered bogus. Some people think that absolute speed is determined by Doppler shift. Maybe Doppler shift and an accelerometer.

[extremecarver]

Have a look at the demo version of GPSAR Pro (GPS Action Replay). This software can trace errors in your tracks (by accelleration, max velocity, etc...) and delete spikes automatically. Works great to smooth out spikes and errors of my tracklogs.

 

Sadly the software is not freeware (it offers much more, but most is specific to GPS Speedsurfing / Speedkiting). The old gpsar freeware version does not offer this feature.

 

see: www.gpsactionreplay.com

 

If anyone knows of any freeware to do this too, please tell here. I only know of GPS Track Analyse.net - which can be used to trace errors, but offers no automatic deletion of errors.

[+user13371]

How does an accelerometer help if you can point the GPS any direction? Don't all of the points have errors and could be considered bogus. Some people think that absolute speed is determined by Doppler shift. Maybe Doppler shift and an accelerometer.

Good question.

 

An accelerometer measures acceleration (surprise). In other words, CHANGES in motion.

 

Let's say you're driving due east at 30 MPH. Taking one fix per second, it doesn't take a doppler shift to figure out your speed (at least roughly) because every subsequent fix is going to be 44 feet dowen the road. A GPS that can factor in a Doppler shift will have an easier time of it.

 

Now suppose the GPS gets a bogus fix, something 100 feet north of where you really are. And maybe the next fix a second later is good, back on the road.

 

If acceleration data is available, the GPS could say to itself "That musta been a big bump to push me off course by a hundred feet, and back again, in just a second or two. But since I didn't *feel* the bump, I think that fix is in error."

 

Of course, a GPS doesn't "think" or "talk" to itself, so that little internal dialog would be conducted in more precise mathematical terms. A good filter would look at fix data, relationships between adjacent track points over time, and measured changes in motion (acceleration).

 

I hope that gives the general idea. Technocrats may now chime in to correct my oversimplifications.

[John E Cache]

If acceleration data is available, the GPS could say to itself "That musta been a big bump to push me off course by a hundred feet, and back again, in just a second or two. But since I didn't *feel* the bump, I think that fix is in error."

OK, now I get what you mean. If the acceleration is 0, you going in a straight line or stopped. I thought you were going to detect turns. I was trying to say if you accelerate the accelerometer won't give you which direction you accelerate because the software doesn't know which direction the GPS is pointed.

[+coggins]

AFAIK, Lee, the PN-40 doesn't use the accelerometer for anything beyond stabilizing the magnetic compass readout.

Bummer, I thought they used it for tighter tracks. Thanks for bringing this topic up Lee.

Edited October 20, 2008 by coggins

[+embra]

I will note that the tracks on the PN-40 have looked very good to my eye...driving around, it looks as if it has the snap-to-road feature (which it doesn't). The track does wander a bit while stationary--the accelerometer could be helpful in dampening that down. Perhaps Delorme can incorporate that data once they get the device out the door and can turn to the fine-tuning.

[+user13371]

... tracks on the PN-40 h...track does wander a bit while stationary--the accelerometer could be helpful in dampening that down.

As an aside, a few years ago when I first got my Magellan SporTrak Pro, I was POSITIVE it had an accelerometer in it.

 

Why? Because when stationary, the display would read "Averaging" - and it was able to detect very small movements (passing the GPS from my right hand into my left) and it would stop averaging for a few seconds. I know now that it DIDN'T have an accelerometer, but I still can't imagine what kind of clever math it took to spot that tiny difference.

[+EScout]

... tracks on the PN-40 h...track does wander a bit while stationary--the accelerometer could be helpful in dampening that down.

As an aside, a few years ago when I first got my Magellan SporTrak Pro, I was POSITIVE it had an accelerometer in it.

 

Why? Because when stationary, the display would read "Averaging" - and it was able to detect very small movements (passing the GPS from my right hand into my left) and it would stop averaging for a few seconds. I know now that it DIDN'T have an accelerometer, but I still can't imagine what kind of clever math it took to spot that tiny difference.

My Meridian does the same. It and my Explorist leave nice tracks. I can hike, stop at a spot, take photos or stay longer, and no extra track points, no spikes. I really like this since using tracks for several uses is my main purpose for GPS.

[+user13371]

My Meridian ... and my Explorist leave nice tracks ...

I felt the same way about my Explorist 210 until very recently, when I carried both it and my eTrex Legend HCx on several walks and bike rides.

 

The Magellan track was smoother. But overall, the Garmin track appeared more accurate when overlaid onto Google Earth imagery. The difference being able to see which side of the road I walked along, exactly where I turned a corner, etc. The Garmin is a bit new to me and I'm still playing with settings to see what works best.

 

Both leave a lot of room for improvement though.

Edited October 21, 2008 by lee_rimar

[+fizzymagic]

The Magellan track was smoother. But overall, the Garmin track appeared more accurate when overlaid onto Google Earth imagery. The difference being able to see which side of the road I walked along, exactly where I turned a corner, etc. The Garmin is a bit new to me and I'm still playing with settings to see what works best.

 

Both leave a lot of room for improvement though.

But you understand that an accelerometer won't fix that, right? I didn't read the whole thread, but I am assuming somebody told you that consumer-grade accelerometers are not sensitive enough by several orders of magnitude and that the stability time is far too short... and that you would need a laser gyro to get the orientation, too?

 

I am guessing that for the kind of inertial navigation you are talking about, current technology could maybe barely do it for > $10K and a box the size of a toaster.

[+user13371]

...an accelerometer won't fix that ... consumer-grade accelerometers are not sensitive enough by several orders of magnitude...stability time is far too short... you would need a laser gyro to get the orientation...

You're over-estimating what I suggested; I'm not talking about anything like a complete INS. Just supplemental data for calculations the GPS does when it "smooths" a recorded track.

 

Even WITHOUT an accelerometer, most GPSRs already do a good job at figuring out your speed and direction (at least above a walking pace). When saving a track, they usually don't save every second-to-second fix; they discard/adjust points to smooth it out.

 

Recording acceleration data (even coarsely defined) would let the GPS look at any points in a series that deviate greatly in speed/direction from adjacent points -- and see if there was a big enough bump to account for it.

[+coggins]

But you understand that an accelerometer won't fix that, right? I didn't read the whole thread, but I am assuming somebody told you that consumer-grade accelerometers are not sensitive enough by several orders of magnitude and that the stability time is far too short... and that you would need a laser gyro to get the orientation, too?

 

I am guessing that for the kind of inertial navigation you are talking about, current technology could maybe barely do it for > $10K and a box the size of a toaster.

I don't know, my kid's Wii remote does a pretty good job with it's ADXL330

[+embra]

Chip (of Team Delorme) made this recent post in the Delorme forum discussing the function of the sensors in the PN-40.

 

Of note for this conversation are these paragraphs:

 

The electronic compass reports information in the Heading info field as well as in the Compass page. If you are stationary or moving slower than 1.5 mph the device will report the heading based on the electronic compass. If you move faster than that it will report a GPS heading since the electronic compass experiences interference when moving.

 

While currently the accelerometer is only used with the electronic compass, we are designing new features and improvements that will take advantage of the accelerometer's ability to detect movement. It will be good to remove the GPS wander when the accelerometer reports that movement is just because the satellites are reporting drift.

 

I'm guessing this thread came to his attention.

[+Team DeLorme]

I'm guessing this thread came to his attention.

Thanks for passing the information along embra, I actually missed this thread. My post on the DeLorme forum was in response to questions from people waiting for their pre-orders to arrive. The accelerometers are very cool, they will definitely help tell the difference between movement and location drift due to poor signal.

 

Chip Noble

Team DeLorme

[+embra]

OK, my guess was amiss. I'm happy for the convergence of thought nonetheless.

[+user13371]

...currently the accelerometer is only used with the electronic compass, we are designing new features and improvements that will take advantage of the accelerometer's ability to detect movement. It will be good to remove the GPS wander when the accelerometer reports that movement is just because the satellites are reporting drift...

Ah, so the answer to the OP isn't a simple, unqualified "yes, the PN40 does that" -- but "yes, eventually the PN40 will do that..."

 

Either way, I'll want one eventually. When will the Macintosh bundle be available?

Edited November 8, 2008 by lee_rimar

[+RRLover]

...The accelerometers ... will definitely help tell the difference between movement and location drift due to poor signal.

Ah, so the answer to the OP is a simple, unqualified "yes, the PN40 does that" -- or "yes, eventually the PN40 will do that?"

 

Either way, I'll want one eventually. When will the Macintosh bundle be available?

It is progressing, DeLorme's working on a lot of things, want to git 'em right, no rushin'.

 

You should see the unannounced PN-40 ZE, (zoolander edition).

 

 

Norm