GPS as a low-speed Speedometer

[Michael GilliganParticipant @michaelgilligan61133]

In a recent thread : **LINK**

https://www.model-engineer.co.uk/forums/postings.asp?th=175373

We considered the use of GPS as a Speedometer for use on a Miniature Railway.

… I continue to have some reservations about its applicability

[ unless, of course, a sophisticated system is installed at the track ]
.

Today, I took my ‘old but respectable’ hand-held GPS [Garmin GPSMap76] for a walk over the hill.

The Trip Computer recorded the following:

Odometer 7.83 km

Stopped time 00:05:18

Moving Average 5.1 kph

Moving Time 01:32:52

Overall Average 4.8 kph

Total Time 01:38:10

Which all seems reasonable, and consistent within itself.

.

There were, however, two oddities :

Max Speed was recorded as 12.0 kph

and for about thirty seconds whilst I was standing on the doorstep, the Speed continued to display 3.0 kph

.

I would be interested to see the results of similar journeys by others.

MichaelG.

[Michael GilliganParticipant @michaelgilligan61133]

[Martin ConnellyParticipant @martinconnelly55370]

The readings are not continuous and the accuracy is not to a fraction of a metre so it doesn't seem odd to me. If you got continuous readings every 0.1 of a second and the accuracy was also 0.1 of a metre you would have a much smoother graph if it was plotted. If you got two readings 0.1 second apart and the position accuracy was 5m too far north for the first one then 5m too far south for the second one it would think you had travelled 10m in 0.1 seconds, equivalent to the 100m sprint in 1 second. When you add the irregular readings to poor positioning the average over time and the distance travelled will be about right but for short times or distances you can expect bizarre results to pop up on the screen.

Martin C

[SillyOldDufferModerator @sillyoldduffer]

Posted by Michael Gilligan on 07/11/2021 14:47:58:

…

.

There were, however, two oddities :

Max Speed was recorded as 12.0 kph

and for about thirty seconds whilst I was standing on the doorstep, the Speed continued to display 3.0 kph

…

GPS is accurate to within a 1m radius with absolutely everything in order and a couple of virgins sacrificed. In practice, accuracy is degraded by weak signals, how many satellites are in view, reflections, solar storms and the quality of the receiver and it's antenna. Various figures quoted, but within a 5 to 10 metre circle is typical. If the receiver can only see 4 satellites, about 15m.

Watching the NEMA output of my GPS module with a antenna looking south through a Window, shows the number of usable satellites seen from my dining room varies between 8 and 16 as the earth rotates, and sometimes drops to 5.

The receiver calculates speed by comparing previous positions. If signal conditions vary whilst stopped, the receiver may detect false movement due to the size of the error circle changing. From the data it appears as if the receiver is moving slowly or jumping rapidly about.

Later GPS and most smart phones incorporate an accelerometer, gyroscope and magnetometer : if present these can detect the GPS is reacting to noise and ignore false signals. I'm not up-to-date with these devices, but I believe quad-copter software has integrated data from all the sensors available for several years. Military for even longer. The mix of position fixing and dead reckoning maintains accuracy in the event one or more sensors are confused. There's no reason why this technology won't end up in a smart phone, hiking system, or car navigation system, or model railway speedometer. Perhaps it already has!

Dave

[Another JohnSParticipant @anotherjohns]

Michael;

A couple of things to think about.

1) Over here (Canada) "WAAS" helps with GPS accuracy, so long as you have line of sight to a satellite in geostationary orbit. Down to about 1.2metres, if I remember correctly, which helps with positional accuracy. Otherwise it's about 8 metres accuracy.

2) I do have a UBLOX GPS receiver with Kaman filtering and (of course) accelerometers; it's used to "smooth out" GPS readings. Unfortunately, you have to do some maneuvers to set it all up; 20km/hr, left and right turns, that kind of stuff. Hard to do that in a sailboat with a displacement hull speed of 6.5 knots, so I do have an accelerometer and have recorded tracks (GPS, with corresponding accelerometer values) to try and roll my own filter; some sailboat owners have found GPS lacking in northern fjords.

3) Heber Valley Railway #618 has/had a GPS speedometer; now way back at 2002 when I was helping, I was informed that the GPS speedo was not accurate below about 5mph. That was before WAAS, etc.

4) You may be near enough to an airport with a GPS augmentation system, but I'm not sure off-hand how normal GPS receivers would find and track this.

All fun stuff – John.

[Clive Brown 1Participant @clivebrown1]

The instrument will calculate speed based on calculated position, which will vary fairly randomly within the margin of error.

The screen display of speed will normally be damped to provide a steady reading.

Michael's 30 seconds of 3 kph might have been due to the loss of satellite cover as he stood in a doorway.

The 12 kph could be the max. instantaneous speed which was damped out and not actually displayed. (Or perhaps he dropped it!)

.

Edited By Clive Brown 1 on 07/11/2021 15:52:31

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Martin Connelly on 07/11/2021 15:16:23:

The readings are not continuous and the accuracy is not to a fraction of a metre so it doesn't seem odd

[…]

but for short times or distances you can expect bizarre results to pop up on the screen.

Martin C

.

Perhaps I didn’t make myself clear, Martin

The reason I was doing this was to test whether or not the GPS system would be suitable for the low-speed application discussed previously

… I thought, and continue to think, that it is not

MichaelG.

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Clive Brown 1 on 07/11/2021 15:51:01:

[…]

(Or perhaps he dropped it!)

.

No he didn’t !

I am incapable of walking at 12 kph, and I was walking fairly steadily.

The 12 kph was most likely due to a brief loss of reception quality … too little to trigger the alert, but enough to diminish the location accuracy.

MichaelG.

[Martin KyteParticipant @martinkyte99762]

iPhones have pretty good accelerometers these days. If you are up for it building an app (unless its already been done) to act as an inertial nav should be acheivable when coupled with the 3 axis magnetometer to provide direction information.

regards Martin

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Michael Gilligan on 07/11/2021 14:47:58:

.

I would be interested to see the results of similar journeys by others.

MichaelG.

.

I am grateful for your comments thus far, gentlemen … but, if it's not too much to ask: Could someone please do me the simple kindness of [approximately] repeating my exercise, using a different GPS.

I have neither an Android 'phone with GPS, nor an iPhone, so I cannot do the comparison myself.

My car GPSs don't have the detailed 'Trip Computer' which the GPSMap76 provides.

I am more interested in the actual slow-speed performance of available/affordable devices than the hypothetical merits of cutting-edge developments.

This is simply a quest for knowledge and understanding.

Thanks

MichaelG

[peak4Participant @peak4]

Michael, I have a variety of devices of varying age; Plain Garmin GPS, Garmin Satnav, and Android phone(s), as well as older stuff such as PDA that I used to use, with built in GPS, and a laptop with a GPS aerial.
I've not found any to be particularly accurate at walking pace speeds.
N.B. My own smartphone apps are all biased towards navigation as opposed to speed measurement, but some of the athletes training apps might well be up to the mark.

I suspect that this is because their sampling rate is optimised for location against battery life, as opposed to speed.
Even the car GPS is slow to start measuring speed when I set off, though it's accurate when moving, but again slow to indicate changes of speed; I'm guessing that is optimised to for processor speed to work best for location/direction/route-finding.

Perhaps this needs looking at from a different angle; GPS speed measurement is used by athletes for training, so it likely that the better units use a much higher sampling rate. Here's a useful link
https://www.sportperformanceanalysis.com/article/gps-in-professional-sports

There are dedicated car/motorcycle speedos available, though not UK legal as the sole speedo in the UK.
https://www.digital-speedos.co.uk/search/gps-speed-sensor

Also available are higher sampling rate GPS receivers to use with conventional digital speedos; these operate at 10-15Hz, which is much faster than conventional hand held GPS navigation devices.
https://dalhems.com/en/p/344ab0b88d444ebfb734a054015bb7a3/GPS-Speedometer-Interface-10-hz-Sampling-Rate

This should give you another rabbit hole to disappear into for the rest of the day.
https://www.researchgate.net/publication/258922748_The_Validity_and_reliability_of_10_Hz_and_15_Hz_GPS_units_for_assessing_athlete_movement_demands

GPS watches have become much cheaper in recent years with non-professional runners to monitor their training, but I've no experience of them.
https://runnersconnect.net/reliable-gps-watches-running/#:~:text=Nielsen%20et%20al.%20reported%20that,error%20margin%20jumps%20to%206.2%25. 

Bill

Edited By peak4 on 07/11/2021 18:09:31

Edited By peak4 on 07/11/2021 18:12:49

[John HaineParticipant @johnhaine32865]

In the other thread I reported some informal experiments with a GPS cyclometer and my phone using a dedicated GPS speedometer and the "GPS Status" app. At least at walking speeds these all gave similar performance, with little sign of erratic readings. I have observed the cyclometer a lot at speeds varying from a few mph up to 20 and above downhill and again it responds smoothly, without erratic readings. Unfortunately neither the cyclometer nor the two apps tested have logging features to directly compare with MG's experiment. The OS app can log routes, I'll see if more detail is available. Interesting to note that neither the cyclometer nor the phone shows any sign of spurious velocity readings when stationary,

GPS does not compute speed by differentiating position w.r.t. time. It has to track doppler shifts of the signals from all the satellites anyway which are moving at a variety of velocities relative to the receiver, so the speed measurement is a by-product of this. Modern GPS has a load of optimisations including for low speed navigation so the latest chips (and the GPS integrated into smartphone chipsets) could be expected to work pretty well. Note also that smartphone GPS makes use of downloaded assistance data. Most modern dedicated "GPS" chips also use 4 satellite constellations, with Glonass, Galileo, and Beidou as well.

I think that a modern GPS, either a cyclometer type or a smartphone running a suitable app, would be very adequate for a model train track speedometer.

Edited By John Haine on 07/11/2021 18:08:01

[Michael GilliganParticipant @michaelgilligan61133]

Posted by peak4 on 07/11/2021 18:05:17:

[…]

This should give you another rabbit hole to disappear into for the rest of the day.
https://www.researchgate.net/publication/258922748_The_Validity_and_reliability_of_10_Hz_and_15_Hz_GPS_units_for_assessing_athlete_movement_demands

.

.

Thanks, Bill … That's one of several papers that I looked yesterday [and realised that I was already being drawn away from my specific area of interest] : Avoiding this 'scope-creep' was why I decided to conduct a simple practical experiment and ask for a little support.

It looks like it's not going to get me anywhere.

Happy to let the thread die peacefully

MichaelG.

[peak4Participant @peak4]

Posted by Michael Gilligan on 07/11/2021 18:32:23:

Thanks, Bill … That's one of several papers that I looked yesterday [and realised that I was already being drawn away from my specific area of interest] : Avoiding this 'scope-creep' was why I decided to conduct a simple practical experiment and ask for a little support.

It looks like it's not going to get me anywhere.

Happy to let the thread die peacefully

MichaelG.

Like you, I've tried, but never had much success with "normal" GPS units for walking pace speed measurements.

Happy to have a play with a smartphone app, but mine is in the middle of 1.2GB download of Maps.me updates at the moment; I might be a while, it keeps failing .

Bill

[John HaineParticipant @johnhaine32865]

If anyone wants some light bedtime reading, this looks like fun:

**LINK**

u-blox was my last employer before I retired…

[SillyOldDufferModerator @sillyoldduffer]

Posted by John Haine on 07/11/2021 18:07:35:

,,,

GPS does not compute speed by differentiating position w.r.t. time. It has to track doppler shifts of the signals from all the satellites anyway which are moving at a variety of velocities relative to the receiver, so the speed measurement is a by-product of this. …

Well, you live and learn! Checking John's statement the first 4 search results all 'confirmed' it was done by calculating d/t. One of the sources is a big name. Then I found this comment on the fifth wrong answer:

GPS receivers do Not measure speed by differencing the positions over time. They could, but it would be grossly inaccurate. The speed is actually measured as a side effect of the solution to another problem: the satellites are moving, causing a Doppler shift in the received radio frequency. The receiver has to adjust for each satellite – but any remaining discrepancy of the received frequency must be due to the movement of the GPS receiver itself. Thereby, the GPS ends up knowing exactly how fast it is moving, to and accuracy much greater than that attainable from position differences. Try searching for GPS Doppler.

Searching for GPS Doppler as suggested by the comment confirms what John says.

Dave

[duncan webster 1Participant @duncanwebster1]

No matter how it works, my Garmin doesn't work under trees especially when it's raining

[Ady1Participant @ady1]

Only the military have access to really accurate GPS

Civilian kit has built in errors and degraded signals to prevent it being weaponised

[John OlsenParticipant @johnolsen79199]

Ady that is no longer true. The feature was called selective availability, it was turned off in May 2000. You can now get the full accuracy that the system is capable of, provided of course that you aren't standing under wet trees.

John

[Michael GilliganParticipant @michaelgilligan61133]

Posted by John Olsen on 08/11/2021 05:54:13:

Ady that is no longer true. The feature was called selective availability, it was turned off in May 2000. You can now get the full accuracy that the system is capable of, provided of course that you aren't standing under wet trees.

John

.

Thanks for putting Ady right on that, John …

As I wrote on the other thread:

.

The basic issues are nicely summarised here: **LINK**

https://www.gps.gov/systems/gps/performance/accuracy/

.

MichaelG.

[Gary WoodingParticipant @garywooding25363]

A little while ago I sent the following email to the authors of a GPS app running on Android.

My wife and I both use Trip Recorder, she on an Alcatel and me on a Moto G. Last week her's registered 3.2 ​miles and mine was 3.1 miles for the same journey. Today​ hers said 3.7 and mine was 3.5, again for the same journey. We start and stop recording at the same time, and walk together. What's wrong please?

They never found a solution.

[Michael GilliganParticipant @michaelgilligan61133]

Thanks for that, Gary … This is just the sort of information I was hoping we might collate.

MichaelG.

[Werner SchleidtParticipant @wernerschleidt45161]

I made tests with my android phone and rfo basic. I can record the readings and all looks good if i drive in my car.

But where i want to use it with my road steamer at slow speeds. I have increments of 1.6 Km/h and that is much worster then a cycle speedometer. On our road are some high trees and there the gps readings are frozen in. It is very useful to record the track profile but not accurate for low speeds. If you want accuracy you have to have a higher update rate and differential gps. And these are not cheap components. The idea with the accelerometers need exact leveling of them, like in a quadrocopter setup. If you use a handheld device,if it is not leveled , that fails by integrating partly the earth acceleration.

For me the only stable solution is a cycle speedometer coupled by wheel impulse.

[Andy StopfordParticipant @andystopford50521]

A few years ago I built a GPS device using a UBlox NEO-6M receiver because I'm hopeless at remembering to write down my work hours. The device used an accelerometer to wake up when moved and record it's position every 10 seconds. I can then enter the data it saved into a Python program which displays it overlaid on an OpenStreetMap map and identify the time I went home.

Now, its important to bear in mind that the NEO-6M is quite old now, and I was using the (probably very cheapo) antenna which came with it, but looking at a track recorded when walking (during Lockdown 1 last year) you can see the track jitters about quite noticeably:

Surprisingly, I can walk in a straighter line than that.

I think with this amount of randomness, a momentary reading of speed is not going to work, though the more data points you can average, the better its going to be (I used a reading every 10 seconds, the largest time interval the NEO-6M allows, to reduce power consumption, and prevent the track file size becoming ridiculously large). The track above is drawn from the raw data points, with no smoothing.

Incidentally, I also tried taking it with me on flying lessons a couple of times – it recorded the track alright (interesting to see how close, or not, the track of one circuit is with another), but it was hopeless with altitude – I believe you need a GPS unit specifically designed for it to do that.

[Michael GilliganParticipant @michaelgilligan61133]

I went for a walk over a different hill today.

As before … the GPS was safely on a neck-strap, it wasn’t dropped, I didn’t catch the ‘bus, etc. etc.

There were, however a small number of “lost satellite reception” glitches.

The results are, again, interesting [to me, at least]

 

The Trip Computer recorded the following:

Odometer 11.43 km

Stopped time 00:50:14 [incudes two visits into shops]

Moving Average 4.6 kph

Moving Time 02:29:57

Overall Average 3.4 kph

Total Time 03:20:10

Which, again, all seems reasonable, and consistent within itself.

.

But this time

Max Speed was recorded as 19.9 kph

and again whilst I was standing on the doorstep, the Speed continued to display 2.something kph until it recognised the loss of reception.

.

Despite John’s assertion to the contrary … I think the incredible max walking speed of 19.9 kph must illustrate that my GPS does rely upon location data for its calculation of speed.

Unless anyone has a better explanation: I can only assume that ‘warp speed’ was engaged during one of the “lost satellite reception” glitches.

… Looking forward to any convincing alternative explanations !

MichaelG.

Edited By Michael Gilligan on 09/11/2021 17:47:04

[Author]

Posts