slmlib

Straight line missions utilities

Objective

Provide a collection of open-source utilities for straight line missions (attempts to cross a geographic area in a straight line) enthusiasts.

Rationale

At this date, the go to online utility seems to be the scoremyline.com website. Given a target line and GPS trace, it provides basic stats (target line length, maximum deviation), GeoWizard's medal ranking (platinum, gold, silver and bronze) as well as a scoring scheme known to the community as the Burdell score (named after the author of that website). However, the mathematics and algorithms involved in such computations are:

1. not particularly well specified
2. not trivial,
3. not open source.

The third point implies in particular that:

1. details and correctness of implementations cannot be verified,
2. community discussions about these de facto standards are difficult to organize.

Disclaimer: we have not been in contact with Mr. Burdell for lack of means to do so (email, social media, etc.). But we would welcome the opportunity.

Implementation

The first efforts are to replicate the results of scoremyline. The bulk of the sample data have been retrieved from the website itself, but also, shout out to the person behind Straight Line Wiki for providing us with additional traces.

Distances and deviations

Measuring distances on earth is a tricky business and it all comes down to the level of precision that you aim for. Thankfully it is an old and very well studied problem. We have implemented the very popular Vincenty inverse problem formula on the WGS84 ellipsoidal model. We chose the level of precision to be accurate to the centimeter compared to more involved methods (well beyond the precision of GPS devices).

Deviations are less trivial to compute as there is no algebraic or numerical solution to the problem on an ellipsoid model. Our implementation assumes (incorrectly) that a straight line on earth (Geodesic) is actually contained in a plane and that this plane contains the center of the spheroid; then it becomes a trivial problem using the algebraic geometry toolbox.

The .sml files issued by scoremyline contain a lot of sample distances and deviations for us to compare and we did not find any discrepancies in computation.

What we did find however is a non-documented choice for track points before the target line's start and after the target line end to be discarded with regards to max. deviation (medal ranking and score), but apparently not when it comes to the leniency setting (what percentage of worst point should be ignored).

The Burdell score

The basic idea of the Burdell score is as follows:

1. divide the target line in segments
2. evaluate the maximum deviation for each segment
3. derive a penalty for the such a deviation
4. aggregate penalties of each segments

Following the description given on scoremyline we manage to be rather consistent with sample values, with few puzzling exceptions though:

Mission	Pro	Pro err.	Amateur	Amateur err.	Newbie	Newbie err.
archie-iom	98.78	-	99.84	-	99.96	-
archie-scotland	90.91	-	99.00	0.04	99.83	0.02
archie-wales-run	98.41	-	99.81	0.01	99.96	0.01
archie-wales-walk	98.32	-	99.80	0.01	99.96	-
geowizard-iom	92.56	-	99.14	0.04	99.83	0.01
geowizard-norway	97.61	-	99.72	-	99.94	-
geowizard-scotland	99.58	-	99.95	-	99.99	-
geowizard-wales1a	0.00	-	0.00	-	44.33	3.70
geowizard-wales1b	63.41	-	95.94	0.15	99.35	0.05
geowizard-wales2	59.82	-	95.76	0.17	99.16	0.01
geowizard-wales3	90.39	-	98.87	-	99.82	0.01
geowizard-wales4	93.41	-	99.30	0.02	99.89	0.01
hiiumaa	0.00	-	84.29	0.70	97.25	0.09
muhu	75.00	0.01	97.10	0.07	99.47	0.02
new-forest	0.00	-	65.84	1.00	94.75	0.19
saaremaa	93.98	-	99.27	0.02	99.84	-
schaffhausen	0.00	-	76.31	0.05	96.04	1.00

It is unclear to us what could explain the differences. More details from Mr. Burdell would probably help. The fact that the errors are bigger as the level (Pro, Amateur, Newbie) decreases and as such length of the segments increases, may indicate that our division logic is different.

How to use it ?

For now there is a library and a CLI tool. You need to compile them

$ cargo build --all --release

The CLI tool

The program takes an input file (CSV or GPX) and optionally the start and end positions and displays the different statistics about about the track.

$ target/release/slm-cli --help
Usage: slm-cli[.exe] [OPTIONS] FILE

Arguments:
  FILE  Input file.

Options:
  -s, --start POINT    Route start point.
  -e, --end POINT      Route end point.
  -f, --format FORMAT  Input file format (default: input file extension).
  -h, --help           Show this message.

Values:
  POINT   Comma separated coordinates (latitude, longitude) as decimal degrees; north and east as
          positive values, south and west as negative values. Ex: '52.606,-1.91787'
  FORMAT  csv: one POINT per line (optional header).
          gpx: first track.

$ target/release/slm-cli fixtures/archie-iom.csv
Route length:             15.0 km
Max. deviation:           21.7 m
Medal rank:               PLATINUM
Burdell score (PRO):      96.0 %
Burdell score (AMATEUR):  99.5 %
Burdell score (NEWBIE):   99.9 %

$ target/debug/slm-cli fixtures/schaffhausen.gpx
Route length:             12.7 km
Max. deviation:           56.5 m
Medal rank:               SILVER
Burdell score (PRO):      0.0 %
Burdell score (AMATEUR):  50.7 %
Burdell score (NEWBIE):   92.4 %