search

esitarski / CrossMgr

Cyclo Cross Management Application
MIT License
40 stars 20 forks source link

'Import rider times from GPX...' wizard #106

Closed kimble4 closed 1 year ago

kimble4 commented 1 year ago

A wizard that reads a GPX track recorded by a rider during the race and attempts to determine their lap times.

Uses the GPX course to determine the lap length and location of the finish, interpolates the recorded track to one point per second, then presents a filtered list of trackpoints and likely crossings of the finish line. These can then be individually selected for import into the rider's lap times (where they will be filtered in the usual manner).

Intended as a backup for when there are problems with a rider's timing tags (or the spotter loses track of their laps), but the rider was recording the race with their GPS receiver. My tests suggest that if you have a high resolution GPX and a reference lap time(s) to adjust the clock offset this can get the times to within a second or two of the RFID system, but it would still be useful to determine the lap count with far worse precision.

Conceivably if you had a lot of patience and trusted riders not to meddle with their recordings you could time a small race this way (as usual, some human input would be required to resolve close finishes), but I wouldn't want to try to manage that at any scale.

esitarski commented 1 year ago

My primary concern is that this is a significant amount of code to write, test, document, support and maintain for a relatively small problem. It also relies on the rider having a GPX device with the ability to produce the GPX file at the race. This reduces the "success factor" even more.

A pragmatic way of handling the "unknown rider" case is to find a "known rider" who was riding "close to" the unknown one. Then:

Not perfect, but you get a working solution with a lot less effort.

To reduce tag failures, it's a good idea to check the tags at race check-in (including using RaceDB, which supports self-serve rider check-in with tags).

On Thu, Mar 2, 2023 at 6:25 PM Kim Wall @.***> wrote:

A wizard that reads a GPX track recorded by a rider during the race and attempts to determine their lap times.

Uses the GPX course to determine the lap length and location of the finish, interpolates the recorded track to one point per second, then presents a filtered list of trackpoints and likely crossings of the finish line. These can then be individually selected for import into the rider's lap times (where they will be filtered in the usual manner).

Intended as a backup for when there are problems with a rider's timing tags (or the spotter loses track of their laps), but the rider was recording the race with their GPS receiver. My tests suggest that if you have a high resolution GPX and a reference lap time(s) to adjust the clock offset this can get the times to within a second or two of the RFID system, but it would still be useful to determine the lap count with far worse precision.

Conceivably if you had a lot of patience and trusted riders not to meddle with their recordings you could time a small race this way (as usual, some human input would be required to resolve close finishes), but I wouldn't want to try to manage that at any scale.

You can view, comment on, or merge this pull request online at:

https://github.com/esitarski/CrossMgr/pull/106 Commit Summary

File Changes

(6 files https://github.com/esitarski/CrossMgr/pull/106/files)

Patch Links:

— Reply to this email directly, view it on GitHub https://github.com/esitarski/CrossMgr/pull/106, or unsubscribe https://github.com/notifications/unsubscribe-auth/AABGXKNWWUMWINCGNYGHYCDW2ET5DANCNFSM6AAAAAAVN7TUXM . You are receiving this because you are subscribed to this thread.Message ID: @.***>

--

Edward Sitarski

kimble4 commented 1 year ago

Fair enough, tbh. This was something I've been wanting to try since a generator failure left us with no tag reads for half an endurance race a couple of years ago. In that case we painstakingly reconstructed the lap count based on a couple of GPX recordings and people's memory of when they lapped who.

(While the generator was quickly restarted, the underlying problem was a combination of a terrible USB-Ethernet dongle that wouldn't bring its interface back up unless the USB side was hotplugged, and a lack of technically competent volunteers to supervise the timing system - they saw a screen full of lap times and didn't recognise the significance of them all being yellow.)

I'm surprised this works as well as it does, but I can see how it would be of marginal benefit for larger scale events where you don't have the problem of the timing team also wanting to be in the race, and where riders lapping each other is a relatively infrequent occurrence (so copying another rider's times is a simple solution).

The dongle has been retired since the purchase of a new laptop with integrated Ethernet, and we've done some work on battery backup for the tag reader, but I'll keep this in my fork as a talisman against future improbable failures...

Kim.

On 07/03/2023 17:58, Edward Sitarski wrote:

My primary concern is that this is a significant amount of code to write, test, document, support and maintain for a relatively small problem. It also relies on the rider having a GPX device with the ability to produce the GPX file at the race. This reduces the "success factor" even more.

A pragmatic way of handling the "unknown rider" case is to find a "known rider" who was riding "close to" the unknown one. Then:

  • Bring up RiderDetail on the "known rider".
  • From RiderDetail, press Edit..., then Copy Rider Times to New Number... When prompted, enter the "unknown rider's" bib number. This duplicates the times of the "known rider" to the "unknown rider".
  • Edit the last duplicated time of the "unknown rider" to get a final finish time (as best as you can).

Not perfect, but you get a working solution with a lot less effort.

To reduce tag failures, it's a good idea to check the tags at race check-in (including using RaceDB, which supports self-serve rider check-in with tags).

On Thu, Mar 2, 2023 at 6:25 PM Kim Wall @.***> wrote:

A wizard that reads a GPX track recorded by a rider during the race and attempts to determine their lap times.

Uses the GPX course to determine the lap length and location of the finish, interpolates the recorded track to one point per second, then presents a filtered list of trackpoints and likely crossings of the finish line. These can then be individually selected for import into the rider's lap times (where they will be filtered in the usual manner).

Intended as a backup for when there are problems with a rider's timing tags (or the spotter loses track of their laps), but the rider was recording the race with their GPS receiver. My tests suggest that if you have a high resolution GPX and a reference lap time(s) to adjust the clock offset this can get the times to within a second or two of the RFID system, but it would still be useful to determine the lap count with far worse precision.

Conceivably if you had a lot of patience and trusted riders not to meddle with their recordings you could time a small race this way (as usual, some human input would be required to resolve close finishes), but I wouldn't want to try to manage that at any scale.

You can view, comment on, or merge this pull request online at:

https://github.com/esitarski/CrossMgr/pull/106 Commit Summary

  • 3fa7dbb

https://github.com/esitarski/CrossMgr/pull/106/commits/3fa7dbbeac66f00bf7d815ff41c7db48377b8564 Initial proof of concept

  • 990ffd2

https://github.com/esitarski/CrossMgr/pull/106/commits/990ffd297e416c18e47e8bcf2cc20b52390945c2 Default offset based on local timezone

  • efb0d7c

https://github.com/esitarski/CrossMgr/pull/106/commits/efb0d7ccc8de957f8b998a2c76d645035c72938c Functionally complete

  • 3412981

https://github.com/esitarski/CrossMgr/pull/106/commits/34129819bb4ef23c6578977abb2a78148f6b295d Add help

  • 62f19ed

https://github.com/esitarski/CrossMgr/pull/106/commits/62f19ed4a5f85c641bd212ddeaf9ead71e37e267 Interpolation

  • e43dd6c

https://github.com/esitarski/CrossMgr/pull/106/commits/e43dd6cfa4715dd482ab13046b9973872e6a7112 Update help

  • 2c2d739

https://github.com/esitarski/CrossMgr/pull/106/commits/2c2d739547d7608138da24079afb16d9baeace6d Remove test filename/bib

  • 8a753d1

https://github.com/esitarski/CrossMgr/pull/106/commits/8a753d11f0d94b5f98c259ac8e46946f9bd37ba9 Tidy up prompt text

  • a97c14b

https://github.com/esitarski/CrossMgr/pull/106/commits/a97c14bccbc753e9e057248e86ff0b7d73679eb8 Merge branch 'esitarski:master' into kimble4-importgpx

  • 63d54ad

https://github.com/esitarski/CrossMgr/pull/106/commits/63d54ad87314ce6f615fc8546fc508cea810b3f4 Error message if no race

  • 93b6c19

https://github.com/esitarski/CrossMgr/pull/106/commits/93b6c1901bbc290c6b587274b6ec3b534435b453 Process lap length, assorted GUI improvements

  • 8a19fd0

https://github.com/esitarski/CrossMgr/pull/106/commits/8a19fd0a1df0666537a91cdc3ffa7530514f3fb0 Update help

  • 8c7931e

https://github.com/esitarski/CrossMgr/pull/106/commits/8c7931e1d7c00539124ca399e22d1284b242877d Fix colouring of lap times

  • d349608

https://github.com/esitarski/CrossMgr/pull/106/commits/d3496083b2560f638bae0e0f781ca510e34e67f9 Merge branch 'esitarski:master' into kimble4-importgpx

File Changes

(6 files https://github.com/esitarski/CrossMgr/pull/106/files)

  • M GpxImport.py

https://github.com/esitarski/CrossMgr/pull/106/files#diff-055ccb8efd9f3821baf221882f1df1493edc54d02c9224618178f98e88d93ca4 (2)

  • M GpxParse.py

https://github.com/esitarski/CrossMgr/pull/106/files#diff-c0c8578fc677080bf5a365698d7f1ce390843425806dc45f4d061d5a4ecd5f9f (2)

  • A GpxTimesImport.py

https://github.com/esitarski/CrossMgr/pull/106/files#diff-41d5ff90a05b50d22700beb0c680458db4cb6de08a700849795bc03242c0ebbc (575)

  • M MainWin.py

https://github.com/esitarski/CrossMgr/pull/106/files#diff-5a5f46554ef499afef996c63f6a0022d0a7e23bfe5636bf5084cff5a35c84db4 (28)

  • M helptxt/Menu-DataMgmt.md

https://github.com/esitarski/CrossMgr/pull/106/files#diff-535776536d0985d0a28ae51b0853436875fd1f857a25c34fa0cc7626bd63125d (12)

  • M requirements.txt

https://github.com/esitarski/CrossMgr/pull/106/files#diff-4d7c51b1efe9043e44439a949dfd92e5827321b34082903477fd04876edb7552 (1)

Patch Links:

— Reply to this email directly, view it on GitHub https://github.com/esitarski/CrossMgr/pull/106, or unsubscribe

https://github.com/notifications/unsubscribe-auth/AABGXKNWWUMWINCGNYGHYCDW2ET5DANCNFSM6AAAAAAVN7TUXM . You are receiving this because you are subscribed to this thread.Message ID: @.***>

--

Edward Sitarski

— Reply to this email directly, view it on GitHub https://github.com/esitarski/CrossMgr/pull/106#issuecomment-1458597400, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHLK6TSZUAPYDNFOPICVRLLW25ZL5ANCNFSM6AAAAAAVN7TUXM. You are receiving this because you authored the thread.Message ID: @.***>