My first HLT use day at AX yesterday. After reviewing the video in detail, accuracy of triggers (start and finish line points) appears to be +/-.25s for the start line and +/-1s for the finish line yielding a timing accuracy of +/-1.25s for each run. This appears to correlate to the timing of the official system for almost every run with one exception where I guess there could have been some mis-trigger of the lights.
However, at AX, if I'm going to use HLT for fine analysis, +/-1.25s is not nearly good enough for viewing run to run differences. How can I improve things?
For every run I appears to have logged 10 fixes per second, but environment was 'Warning' due to magnetic conditions. The accuracy graph looks to be very good - small circles.
Thoughts appreciated.
db
Trigger location consistency with XGPS160
Re: Trigger location consistency with XGPS160
The environmental warnings will generate a GPS drift. This means there is a certain offset between real position and measured position beyond the reported accuracy. As long as trigger definitions and actually driving within a short time frame, that's no problem. This shift changes over time. So in case you have an hour or more in-between, your recorded with be shifted and measurements can get wrong. The named timing accuracy is probably due to some other issue. For a 10 Hz device showing good accuracy, timing will be within 0.05s mostly. A bigger difference means the position used for start and stop are different to the positions the official system measured.
- Harry
- Harry
-
- 20 or more Posts ★★★
- Posts: 836
- Joined: Thu May 03, 2012 5:26 am
- Location: Kingsport, TN USA
Re: Trigger location consistency with XGPS160
Be sure the XGPS160 has acquired lots of satellites before setting the POI's. In GPS view, there's a progress bar near the top of the page with the number of satellites acquired below it. Wait until the indicator is all the way to the right. It takes a few minutes after the XGPS160 gets its initial fix, i.e. the led stops blinking.
Re: Trigger location consistency with XGPS160
I have done further analysis and now realized a big mistake in my original methodology. I was using the video immediately viewable within HLT as my reference and assuming constant offset from start and stop triggers - this was a mistake.
I have now overlaid and exported 5 laps with good data from the afternoon.
First, I compared clock time to pass the first clear visual reference after the start. This is just over 2 seconds in, but is SPEED DEPENDANT, so some variation is likely.
Lap, Time to First Marker
1 2.31
2 2.52
3 2.28
4 2.64
5 2.38
Given likely variations in speed and limited frame to frame resolution, I'd say there was likely very little deviation from a constant starting point.
Finish times gives a better opportunity as they are not dependant on speed but on on errors introduced by a start point offset. For this test I logged the difference between finish time and time passing a visual marker near the finish (I have no way to know exactly where the HLT GPS finish is on the video):
Lap,Finish Time,Finish Marker,Delta
1 69.30 69.27 -0.03
2 66.45 66.39 -0.06
3 65.85 65.81 -0.04
4 65.04 65.00 -0.04
5 64.28 64.24 -0.04
So if we assume that the finish marker is 0.04s at speed away from the finish trigger there is only +/-0.02s variation and that includes variation introduced at the start trigger. I'd say that is impressive!
I hope to have official times soon and that will also indeed be interesting!
db
I have now overlaid and exported 5 laps with good data from the afternoon.
First, I compared clock time to pass the first clear visual reference after the start. This is just over 2 seconds in, but is SPEED DEPENDANT, so some variation is likely.
Lap, Time to First Marker
1 2.31
2 2.52
3 2.28
4 2.64
5 2.38
Given likely variations in speed and limited frame to frame resolution, I'd say there was likely very little deviation from a constant starting point.
Finish times gives a better opportunity as they are not dependant on speed but on on errors introduced by a start point offset. For this test I logged the difference between finish time and time passing a visual marker near the finish (I have no way to know exactly where the HLT GPS finish is on the video):
Lap,Finish Time,Finish Marker,Delta
1 69.30 69.27 -0.03
2 66.45 66.39 -0.06
3 65.85 65.81 -0.04
4 65.04 65.00 -0.04
5 64.28 64.24 -0.04
So if we assume that the finish marker is 0.04s at speed away from the finish trigger there is only +/-0.02s variation and that includes variation introduced at the start trigger. I'd say that is impressive!
I hope to have official times soon and that will also indeed be interesting!
db