Distance Measurement Discrepancies Explained. Remote GPS antenna/receiver for smartphones running mapping and timing apps

Typical spatial sensitivity plot of a smartphone GPS receiver chip.

• This technical essay discusses and seeks to address the two known issues that limit the accuracy and precision of GPS devices when used in motorcycle backcountry navigation and rally timing. These issues are the discrepancies observed measuring distance travelled and the occasional non-detection of virtual timing gates by GPS location.

• These two issues are related but not direct dependent on each other.

• Methodological and equipment solutions are discussed.

Introduction

If you have followed this blog or participated in a VIME event you will be familiar with the conversations around the apparent (real in many instances) mismatch between distance measurements between two geographic points and the GIS (Geographic Information System) predicted distance as produced by Rally Navigator (the computer program used to write roadbooks) when measured by GPS odometers. Also there are instances when smartphones fail to detect their presence at GPS locations that correspond to rally timing points.

Sources of discrepancies.

Car regularity (TSD) rallies are famous for the precision that road distances are measured to. To be fair, the precision achieved is a combination of highly calibrated odometer instruments and methodological techniques that do not easily transfer to motorcycles.

To expand on this, car TSD rallies are set using a precise, wheel-driven odometer that regardless of it’s variation to the actual, on-the-road distance, is used as the “master measurement”. Distances between tulips are measured with this “master odometer” and are written into the roadbook as the “event gold standard”. To enable competitors to calibrate their odometers (usually wheel driven) to the event master measurement, a calibration leg is set at the start of each route that allows each car navigator to observe the difference between their odometer measurement and the master measurement. This is then quickly converted into an “adjustment factor” that each leg distance needs to be multiplied by, in-car and on-the-fly to convert their measurements to the master measurements.

The amount of work required by event course setters is quite considerable as master measurements need to be written into the roadbook during the editing and proof reading process (an adjustment that requires some “track manipulation” in Rally Navigator), the exact path on the road needs to be apparent to crews and of course copious notes need to be made by the course-setting crew.

The first two sources of error for motorcyclists….

The rally motorclist is constrained in their choice of odometer to either the wheel driven odometer that may be a secondary add-on device, or part of the machine’s speedometer (which can be notoriously wayward once the statutory inaccuracy is accounted for - 10%+4kph above the true speed but no lower than the true speed.

Add in changes to the tyre size used, deep knobbly tyres wearing, tyre pressure changes and “tangential route” errors it is not surprising the speedo odometer can only be partially relied on.

OR

The amateur motorcycle Rallyista is mostly confined to using a GPS driven odometer - either as a function of the roadbook PDF reader of choice or a standalone device.

The workload of the rider would probably exceed manageable limits if a “calibration leg” were to be provided for riders to calculate their odometer calibration factor and have to multiply every distance in the roadbook by this number AND keep a running tally of the overall distance travelled in addition to adjusting the control point to control point distances.

Rally Roadbook Reader GPS odometer “offset adjustment”. This smartphone app has an odometer calibration offset adjustment in the settings, up to +/- 5%. On reflection, the intended purpose of this feature is to permit “synchronisation” of the GPS odometer to the “master odometer” used to set car TSD road rally courses. If you remember, tarmac rallies have the distances between waypoint tulips measured “over the ground” rather than relying on trigonometric GIS calculated, horizontal displacement distances.

The next source of error comes about from the GIS distance measurements made by Rally Navigator. The GIS system calculates the distance between the map points entered along the proposed route. The computer system is not able to automatically plot a smooth course on the ill-defined gravel roads, so you get very precise trigonometry-derived measurements of somewhat poorly positioned geographic locations. This inaccuracy can be mitigated by placing strategic control points at unambiguous locations such as bridges over rivers, major road junctions and immovable geologic features. These points can then be used to reset the route odometer to the roadbook and go again.

Tangential and Orographic errors.

In addition to curved routes plotted in Rally Navigator being rendered as a series of straight lines (and so introducing a distance error), the path ridden by a motorcyclist is likely to differ from the path plotted (on the GIS map) in Rally Navigator and the median route of the road.

The mapped world, it would appear is made up of a series of straight lines while the ridden world is made of beautiful curves…

Orographic? “Orography is the study of the topography of mountains, and can more broadly include hills, and any part of a region's elevated terrain.”

Simplistically, maps are a two dimensional representation of a three dimensional landscape. The ultimate bird’s eye view. There is an unavoidable distortion of distance when representing steep ground on a map.

To explain, taking a bird’s eye view of a road on flat terrain, from the high mapmaker’s viewpoint, the over-the-ground distance along that road is as the mapmaker sees it.

Now compare that situation to an extreme Loonie Tunes world where the roadway then goes straight up a vertical cliff face (of say 1000metres length). In this make believe situation, the roadway then goes would be 1km long, straight up. But from the mapmaker’s extreme altitude viewpoint directly over this cliff, the road would have no distance at all. Just a line on the map.

While this cartoon example is absurd, reality lies between the horizontal and vertical and demonstrates how the projected map distance of a road going up a hill will always be less than the over-the-road distance.

Smartphone GPS receiver chip limitations.

The image at the start of this essay indicates the spatial sensitivity of the GPS chips typically found in smartphones.

To back up a little, all things GPS rely on the antenna-receiver picking up the radio signals emitted by the GPS satellites in orbit around the planet (no, it’s not a flat Earth!). They are not strong signals and anything that obscures the view of the sky will interfere with the GPS signal. Tall buildings, mountains, trees, leaves, the orientation of the receiver chip to the sky…

Typically, smartphone GPS receivers are accurate to a 10m (30foot) diameter circle under open sky…. Introduce obstructions and this “circle of uncertainty” will increase.

Smartphones also often receive GPS signals on one radio frequency and update at a rate of once per second (1Hz) and draw positioning data from the US NAVSTAR GPS satellite constellation.

Take another look at the colour image at the start of this essay, the colour coding helps to demonstrate how the receiver sensitivity varies with orientation.

The receiver is most sensitive in the “up” direction when the phone is held flat as if it were on a tabletop.

The receiver is most insensitive when it is held vertically as in a pocket.

Now think how most people carry their phones…. Either near vertically on a navigation tower or vertically in an inside pocket.

Combining all these challenges to smartphone GPS performance, it is surprising to me that the GPS works as well as it does.

Nav tower orientation for smartphones - not the best for GPS sensitivity

So here is a worked out problem….. if the GPS “circle of uncertainty” (to borrow from Mr. Heisenberg) is 10metres across and the GPS sampling rate is 1 Hertz, how far would a motorcycle travelling at 60Kph travel in one second?

60Kph = ( 60/(60x60) ) x 1000 meters per second. Using BODMAS this simplifies to

1/60 x 1000 m/s =16.667 m/s…

At 60Kph, a bike would cover 16 2/3 metres in a second - and the circle of uncertainty could be 10 metres. On the face of it, the bike and rider could come and go before the GPS had a chance to register the fact.

Fortunately the builders of the GPS timing system used (Richta) realised this and set the detection radius of control points to be rather larger than the “circle of uncertainty”.

Nevertheless, if the view of the sky is hampered by trees and foliage, and the phone has been out of contact with the GPS satellites for a long enough time to require satellite signal reaquisition, that GPS control point may go unrecorded through no fault of the rider or the course setter…

Glossary:

GPS = Global Positioning System

GNSS = Global Navigation Satellite System

GLONASS = The Russian GPS satellite constellation. 24 satellites

Galileo = The European satellite constellation. 30 satellites

BeiDou = The Chinese satellite constellation. ?35 satellites

NAVSTAR = The US satellite constellation. 32 satellites

So we have these following points to bear in mind…

• Assume the GIS (Geographic Information System) trigonometrically calculated distances between points to be always correct - but only when measured by GPS odometers.

• Also assume the GPS measurement of distance travelled (and implicitly, speed) is also always correct, but only when used against GIS generated distance measurements.

• The course plotted in the roadbook writing computer system won’t always exactly match the exact route of the road, and

• The route followed by a motorcyclist won’t always exactly match either the median route of the road or the route plotted in the roadbook.

• There isn’t an easy way to verify and edit into roadbooks “master odometer” distance measurements made when verifying and pre-riding routes.

• The odometers on motorcycles, regardless of type are always liable to differ from the roadbook distance measurements, although it is difficult to say what measurement is righter or wronged than others…

• Take steps to minimise the challenges your smartphone has in finding GPS satellites from which to take data

This has been a rather long journey to arrive at the sources of GPS variances, disagreements, deviations - call them what you will. But what is to be done about them?

Taking them in order…

1. In the absence of a “super odometer” for the motorcycle rally setter to use (remember that lucky person has the same equipment as event competitors), the trigonometry calculated distances between waypoints has to be taken as gospel - EVEN THOUGH IT ISN’T!

   GIS computer systems calculate the horizontal distance between two marked points, not the on-road distance. GPS odometers also calculate horizontal distance, wheel-driven odometers measure on-road distance.

   There is no easy/cheap/swift way of doing distance measurements differently or better. Rally setters need to let it be known how they measure distance - GPS or wheel, horizontal displacement or on-road. Rally-setters and participants can develop an empirical technique of noting roadbook distances at strategic, incontrovertible geographic locations and resetting their trip odometer to the roadbook distances without troubling themselves with questions about whose measurement is correct.

   *VIME events will be measured using the GPS/horizontal displacement method.*

2. Within the instrumentation and processing limitations of smartphones, their distance-measuring accuracy has to be taken as incontrovertible. SMARTPHONE ODOMETERS MEASURE HORIZONTAL DISPLACEMENT, NOT ON-ROAD DISTANCE.

   There are steps that can bet taken to improve the sampling rate and location precision of smartphone GPS systems, more on this later (GPS remote antenna/receivers and specifically the SkyPro XGPS160).

3. The course plotted will always be a series of short straight lines and the road on the ground may differ in location and route to that shown on the map.

4. Ride a bike over a rough road and you will choose the lines of least resistance. The distance discrepancies you introduce will be added to the overall discrepancy. Some will cancel out, others will add.

5. While it’s not strictly true to say that control-point to control point distances as measured in Rally Navigator can’t be adjusted (extra path diversions or corner shortcuts can be introduced in order to massage leg lengths up or down), this editing would require every leg length to be measured, the measurement would be subject to every source of variation discussed here and every “intense” competitor would need to calibrate their odometer and multiply every roadbook distance by their “adjustment factor” - as mental arithmetic. This would create more angst among riders than acknowledging the measurement shortcomings and using empirical workarounds.

6. This is a methodological difficulty covered in point 2 above. Horizontal displacement versus on-road distance, GPS versus wheel driven. Two different measurement systems not to be combined if reliable results are desired!

7. This is the “if you didn’t understand all the foregoing, just do this” bit…

   If you’re on an android, set it to airplane mode so it can concentrate on your needs.

   Carry your Richta-running phone and your roadbook PDF-reading phone as close to horizontal as possible.

   Reset your odo (whether it is wheel or GPS driven) to each “waypoint of significance” distance at every opportunity.

   Consider using a remote GPS receiver-antenna. This is the point of this essay.

   
   
   
   

Dual Electronics Corp. SkyPro XGPS 160

After much consideration, comparing the different features and capabilities of different receiver-antennas and reading about which model other rally enthusiasts had chosen, I bit the bullet and bought one of these.

$230 plus all the usual hidden costs, I bought mine from Amazon rather than traipsing around the gadget shops.

The plus points:

• It receives GPS data from both the US GPS satellite constellation and the Russian GLONASS system simultaneously.

• It can feed that information to upto five devices simultaneously by Bluetooth.

• It can operate while being charged from a USB charger.

• iPhones can take Bluetooth data from more than one device at a time - so you can use a Bluetooth pdf controller to scroll your roadbook at the same time that the XGPS160 is feeding it with GPS data.

• It pairs with iPhones really easily

• It can also talk to Android phones although the pairing process is more complicated.

• It updates GPS data a a rate of up to 10 position samples per second (ten times faster than phones typically).

The minus points:

• It is fairly bulky, 70mm x 55mm x 22mm and the sticky holder is 110mm x 120mm x 25mm

• It has to be mounted horizontally, facing straight up.

• It isn’t rainproof - it is intended to be mounted on the dashboard of a car or light aircraft under the windscreen.

• Some ingenuity is required to successfully mount it to a rally motorcycle used in the harsh conditions that can be expected.

What it won’t do:

• It won’t get your GPS to work in impossible situations, like in skyscraper urban environments, in tunnels.

• It won’t solve any of the distance measuring problems caused by the issues described above.

What it should do:

• Higher frequency data sampling and drawing data from two satellite systems simultaneously should minimise missed GPS control points and “uncounted kilometres travelled” due to the GPS reception dropping out.

Does it do it?

• I shall test it and let you know. I suspect I am not doing anything new or unique here, these little buggers have been used in the car communities for quite a number of years now but to my knowledge not transferred to a motorcycle environment.

This is how I have fitted the XGPS 160 to my bike…

The XGPS 160 does what its makers say, it can talk to upto five devices by Bluetooth while being powered from an external USB. iPhones can receive Bluetooth from (at least) two devices simultaneously.

The closing message is this.

• There isn’t a perfect solution for our distance measuring difficulties.

• They arise from the combination of methods, systems and devices we use in the steep environments characteristic of VIME events.

• These combinations cause distance measuring challenges that can’t easily or cheaply be overcome.

So

We continue to live, ride and play in a crazy, mixed-up rock and roll world using methods and instruments with known flaws to achieve the improbable. Beyond all of the above, there is a limit to the accuracy and precision our efforts can achieve. In the light of all these flawed and sometimes mismatched measurement systems we have ways of getting the best out of them and it’s probably best not to take ourselves too seriously ;)

See you at the next one.

JDB

FURTHER EDIT…

With the agglomeration of all the marginal discrepancies outlined here AND the suggested mitigation strategy of resetting the odometer at every waypoint of significance, there is little point in adopting the car regularity rally customer of having an odometer calibration leg at the start of each route as the discrepancy between different odometers will be manifest as soon as the first route “challenge” is encountered (lost satellite signal, road incline, route tangent, difference between GIS and actual road placement). An odometer calibration leg would not add anything to distance measurement accuracy.

EDIT

There is yet another source of GPS odometer discrepancy that I have not discussed.

A GPS odometer measures distance travelled using the location data breadcrumb trail it builds up from the GPS locations it collects.

If the antenna-receiver should loose connection with the satellite constellation it is using, the device won’t have any information about the route it was taken on while it was off-line. It has already been noted that foliage cover obscuring the view of the sky will be sufficient to blind the GPS.

When satellite communications are re-established, the odometer will metaphorically draw a straight line between the two known positions and assume that was the route taken.

If the actual route was anything but a straight line, there would be a clear difference between the mapped (twisty) distance and the assumed straight GPS distance.

If a remote GPS receiver/antenna decreases the likelihood of satellite communication being lost, this source of discrepancy will be eliminated or reduced. In signal processing, this is called an aliasing artefact. See https://en.wikipedia.org/wiki/Aliasing

Showing the effect of under-sampling an input signal when reconstructing to give a result that bears little resemblance to the original.