How exact is the velocity that is showing in a GPS Navigation System like Navigon or TomTom?
2 Answers
This depends upon a large number of variables that are imposed by the GPS navigation system in use and by the processing of the data by the SAT-NAV following the determination of the position fix. These include
- The interval between position fix solutions. This is normally 1 second.
- The position accuracy of each of the positions used to determine the velocity. This depends upon the quality of the sky view and, to a certain extent, the length of time that the GPS has had a sky view.
- The number of positions used to calculate the velocity
- The acceleration of the unit
- Filtering algorithm used to smooth the velocity value
The velocity value from the GPS receiver is calculated from the difference between position fixes. As these are typically generated at a 1Hz rate you are measuring a distance travelled in 1 second. The position fixes can have an error of a number of metres, with successive positions wandering from the true position, sometimes by a metre or more for a static unit. I have had static units using high end GPS chipsets that have apparently had a velocity of up to 2 m/s while fixed to my office building.
Some units may filter the velocity value or use more than 2 position fixes to calculate the antenna velocity. This will add a lag to the velocity value shown as the antenna accelerates.
Providing a definitive answer is not possible in just the same way that defining position accuracy has to be bounded by many caveats around the sky view and number of satellites visible.
-
\$\begingroup\$ Thanks a lot for your excellent explanation! I know that providing a definitive answer is not possible. Was i was wondering if the tollerance is more like +/-1% or +/- 10%. \$\endgroup\$– gsharpMar 30, 2011 at 12:06
-
\$\begingroup\$ Tolerance is not going to be a % of actual value. My static unit can move at 1m/s (sometimes 20m/s vertically :-) !!) which is an infinite % error. \$\endgroup\$– uɐɪMar 30, 2011 at 12:14
With no filter at all the most error you should ever see is 30 meters/second, but this will only last for 1 sample. In my past experiences, you should expect to see an error more around less than 1 meter a second, but with 0 mean. This equates to about +/- 2.25 miles per hour. However, for consumer GPS modules you will never see an error like this. There are several reasons for this, but it all comes down to filtering.
GPS systems like the Garmin and Tom Tom employ filtering algorithms to give you the most accurate position it can. One of the things that these systems do is they assume you stay on the road until they are very sure that you didn't. With the filtering on the position, it forces the velocity calculations to already be much more accurate then using the raw data. Then with the filtering on the velocity calculations you get a very accurate measurement.
NOTE: I should probably quantify what I mean by "very accurate". I thinking about the situation of you having a GPS in a car and comparing its accuracy to what the car says it is at.
-
\$\begingroup\$ Filtering slows down the response though, so if you are accelerating, your car might show 50 mph and the GPS 45 mph, so that's more than your 2.25 mph error. It'll catch up within a second or two though. \$\endgroup\$– davrMar 30, 2011 at 17:55
-
\$\begingroup\$ @davr correct, and if someone really needed more accurate data they could add an accelerometer to fill in the gaps and get faster response. \$\endgroup\$– KellenjbMar 30, 2011 at 22:18