GPS Logging accuracy - why the lies?
Discussion
I've been having a chat with a few people about data logging and it got me wondering why do so many apps / products show far greater resolution than their receivers are actually capable of?
Many android / apple apps show tenths of a second and some even show hundredths of a second when nearly all mobile phone GPS chips operate at 1hz.
When it comes to things like the Racelogic vbox they also show hundredths of a second when they only have a 10hz chip. Their manual even shows "Resolution 0.01s / Accuracy 0.1s". Why the trickery?
Are they doing something clever?
Do people just want to see more numbers even if they are meaningless?
I suppose if they utilise G sensors and they know that a vehicle is still accelerating they can use that overlapped with the GPS data so if it knows at 10.1 second the car is doing 59.5mph and it's still accelerating at a rate of xG it'll hit 60mph before the next GPS refresh cycle. - maybe I've just answered my own question whilst typing. I suppose it knows when each cycle is so it can tell that if it's gone above the vale by the next cycle it can work out the acceleration between the cycles and it's only simple math to work out roughly when it'd be at a certain point.
I'm just hoping that some of you out there have wondered the same in the past and have an answer.
Many android / apple apps show tenths of a second and some even show hundredths of a second when nearly all mobile phone GPS chips operate at 1hz.
When it comes to things like the Racelogic vbox they also show hundredths of a second when they only have a 10hz chip. Their manual even shows "Resolution 0.01s / Accuracy 0.1s". Why the trickery?
Are they doing something clever?
Do people just want to see more numbers even if they are meaningless?
I suppose if they utilise G sensors and they know that a vehicle is still accelerating they can use that overlapped with the GPS data so if it knows at 10.1 second the car is doing 59.5mph and it's still accelerating at a rate of xG it'll hit 60mph before the next GPS refresh cycle. - maybe I've just answered my own question whilst typing. I suppose it knows when each cycle is so it can tell that if it's gone above the vale by the next cycle it can work out the acceleration between the cycles and it's only simple math to work out roughly when it'd be at a certain point.
I'm just hoping that some of you out there have wondered the same in the past and have an answer.
Edited by ChrisC-Berks on Tuesday 3rd July 09:31
Sea Demon said:
Maybe they know more about it than you do? Do you work in the industry?
I would hope they do know more than me. I understand the simple concept of reading logs and the fact they show a higher resolution than accuracy seems daft. They are accepting that the resolution isn't exact.Car-Matt said:
Interpolation
This seems to be the most obvious answer. I think I pretty much wrote this in far more words when I was trying to work it out for myself.It's pretty much as you wrote (interpolation).
The lower end Racelogic stuff at 5 or 10 Hz (their expensive stuff is 100Hz) seems to give fairly good numbers when interpolating for calculations due to their algorithms, kalman filters etc.
Some users are obsessed with having another number after the decimal point even if it's meaningless.
I've had the conversation numerous times about all kinds of measurements and it's bizarre when people want something to 4 significant figures (e.g. 55.32 bar) when the actual uncertainty (see ISO17025 for a thrilling read) could easily be 50 times the value of that 4th significant figure. i.e. the actual value is 55.3 bar +/- 0.5 bar.
The lower end Racelogic stuff at 5 or 10 Hz (their expensive stuff is 100Hz) seems to give fairly good numbers when interpolating for calculations due to their algorithms, kalman filters etc.
Some users are obsessed with having another number after the decimal point even if it's meaningless.
I've had the conversation numerous times about all kinds of measurements and it's bizarre when people want something to 4 significant figures (e.g. 55.32 bar) when the actual uncertainty (see ISO17025 for a thrilling read) could easily be 50 times the value of that 4th significant figure. i.e. the actual value is 55.3 bar +/- 0.5 bar.
You'd have to be doing some serious trickery to get meaningful accuracy out of a raw GPS at that scale regardless of the update rate and number of channels it could handle.
The heavy lifting - whether it's for a datalogger or something more serious - usually falls onto inertial measurement i.e. accelerometer + gyro.
GPS is more for initialisation (on an aircraft you use the GPS to tell the inertial system where it is when it powers up) and position comparison.
The other nice thing the GPS signal gives you is a solid time base for your other recorded data. The calculations for everything only really work if your timing is reliable.
Proper car measurements would use something more reliable like beacons and laser ground sensors.
The heavy lifting - whether it's for a datalogger or something more serious - usually falls onto inertial measurement i.e. accelerometer + gyro.
GPS is more for initialisation (on an aircraft you use the GPS to tell the inertial system where it is when it powers up) and position comparison.
The other nice thing the GPS signal gives you is a solid time base for your other recorded data. The calculations for everything only really work if your timing is reliable.
Proper car measurements would use something more reliable like beacons and laser ground sensors.
Gassing Station | In-Car Electronics | Top of Page | What's New | My Stuff