My GPSMAP 67 Perfect Track

[mimichris]

The track of my data from yesterday, the track is perfect, one point every second.

Spoiler

Eucalyptus-2023-05-15 16.28.47.rar

[Nail]

Did you have a 66sr with you?

[mimichris]

No. I didn't have the 66sr.

[Nail]

Are you planning such a test of two devices on a walk?

[mimichris]

I just did 5.5km, with the 66sr and the 67, I took one second for the recording interval, in fact it's ideal, the two tracks overlap perfectly, nothing to say.

Spoiler

66sr-2023-05-17 15.50.36.rar

Spoiler

67-2023-05-17 15.49.46.rar

[GPSrChive]

So you are happy with the 67 accuracy?

[Nail]

I just did 5.5km, with the 66sr and the 67, I took one second for the recording interval, in fact it's ideal, the two tracks overlap perfectly, nothing to say.

Track 67 is not perfect. The 67 does not keep a straight track compared to the 66sr.
I had the same when I took two devices, 67 was like after three beers.
The 67 has a lot of other advantages, but the 66r is better here. Garmin still needs to work on accuracy a bit. I hope it will be soon, then I will go back to 67.

Spoiler

67vs66sr30.jpg

[MarkHL]

I just did 5.5km, with the 66sr and the 67, I took one second for the recording interval, in fact it's ideal, the two tracks overlap perfectly, nothing to say.

Track 67 is not perfect. The 67 does not keep a straight track compared to the 66sr.
I had the same when I took two devices, 67 was like after three beers.
The 67 has a lot of other advantages, but the 66r is better here. Garmin still needs to work on accuracy a bit. I hope it will be soon, then I will go back to 67.

Spoiler

67vs66sr30.jpg

I'm not sure the 67 is a less accurate instrument, only that it lays a less detailed track. The issue may not be the "point" accuracy but a how the software interprets the point. From previous posts, it appears the 67 has improved nesting interpretation and doesn't wander from a static location to the extent that previous models have. So, the software may be excluding points that it has low confidence in. The same thing may be occurring when it represents a track on the device as a straight line versus a more detailed path. It may not have confidence in a particular point that if used may have given the track more detail. So previous models may have represented a more detailed track, but not necessary more accurate.

Bottom line is; accuracy is really where a point is laid and not the final track.

Plus, the device does not lay the track in real time, but only updates it intermittently. Often the track being laid is represented on the device as a straight line and after a period of time, it updates the track with more detail. It's now using the points it deems as accurate to update this track and not using the ones it has less confidence in.

It's a bit of a compromise with the software determining which points to used and which points to exclude. It's this logic that may be giving the 67 an appearance of being less accurate when it's really an evolution of a more refined way of using the data.

Is this logic too restrictive? Possibly, and this may be where the improvements can be made.

[Nail]

It's a bit of a compromise with the software determining which points to used and which points to exclude. It's this logic that may be giving the 67 an appearance of being less accurate when it's really an evolution of a more refined way of using the data.

A very interesting statement. I also had this belief that 67 would be more accurate than 66sr, and it probably is. The problem is that between the exact points there is a padding with points that are not exact. When you look at the gps track, it undulates evenly. This may be due to shutting down certain processes to save energy. The key will be the word you used "compromise".

Is this logic too restrictive? Possibly, and this may be where the improvements can be made.

I look forward to these improvements.
Mam nadzieję, że jest na to przestrzeń.

[Nail]

I'm not sure the 67 is a less accurate instrument, only that it lays a less detailed track. The issue may not be the "point" accuracy but a how the software interprets the point. From previous posts, it appears the 67 has improved nesting interpretation and doesn't wander from a static location to the extent that previous models have. So, the software may be excluding points that it has low confidence in. The same thing may be occurring when it represents a track on the device as a straight line versus a more detailed path. It may not have confidence in a particular point that if used may have given the track more detail. So previous models may have represented a more detailed track, but not necessary more accurate.

Today I checked the shared tracks again and there is a big difference in the number of points (66sr 3948) vs (67 4569) 16% more has GPSMAP 67. Your explanation is true for 66sr, not 67. 67 fills the track with randomly placed points, as for of which he is not sure and corrects from time to time to the right point. Therefore, 67 is still within the assumed error corridor. The track GPSMAP 67 is not detailed or accurate, more added points do not confirm it at all. I still believe that processes 67 do not run continuously.