mendhak/gpslogger

Filter on stability : set a traveled distance threshold

Opened this issue · 4 comments

Hi Mendhak,
Great work !
I think another filter should be interesting :
As the level of accuracy is reached (for example EPE=3m)
the distance traveled recorded through later points continues to grow,
firstly by some meters, then some decimeters, and finally, after 10 or more measures,
the distance is stabilized within 3 or 5 cm.
I think it should be interesting to set another optional filter on this variation of distance traveled to obtain more precise records.
What do you think ?

The aim is to record numerous measures on a fixed point, for example 24 records a day.
To process them and try to obtain a more accurate position.
But to enhance these measures it would be necessary that each record be as precise as possible, not a bare record of the first measure available for the "accuracy" filter set up.

Hi Mendhak,
No belief on that option ?

Hi I've had some time to think over this properly, it is too complex to implement and maintain. There would be a lot of ongoing data to hold to perform the accuracy changes, and also to implement a backoff when the user enters buildings where there is bad accuracy.

Accuracy has been a weak point in GPSLogger for a long time. I think this could be better managed by a long requested change which is, instead of just taking the first match that comes along, let the app run for a certain duration and choose the best match in that duration. That's also somewhat complex to implement and introduces another confusing option, but is more general purpose.

The aim is to record numerous measures on a fixed point, for example 24 records a day. To process them and try to obtain a more accurate position. But to enhance these measures it would be necessary that each record be as precise as possible, not a bare record of the first measure available for the "accuracy" filter set up.

I think it would defeat the purpose of GPS logger. It is made and designed to record movement, not to improve accuracy.
For example, it has no way of knowing if the device has moved or if it's still in the same place. It should not try to wear two hats on the same day.