Pong

## June15th 2015 GPS Hickups

After fitting the GPS measurements to a parabolic curve, the jitter has been reduced significantly. Still, when the GPS is starting up, the device may report locations that are way off, at least for the first couple of measurements. Even if the device is stationary, the measurements will be wandering around. These two categories of false measurements need to be suppressed.

Fortunately, the GPS device is pretty good at estimating the velocity. The velocity estimates are usually much better than the positional estimates, since the device is utilizing accelerometers to find out whether the device is moving or not.

So even though the reported locations might be showing a movement, examining the reported velocities indicates that there actually cannot be such movement, if the velocities are close to zero.

More precisely, if the reported velocities $v_i$ are small compared to the measuring accuracy $a_i$, we can safely deduct, that the device is not moving at all and the movement is just jitter. In such a case we simply suppress all positional measurements:

if $\frac{ min(v_i, v_{max}) }{ a_i } < \frac{ v_{max} }{ a_{max} }$ → suppress measurement

This means that points $p_i$ with accuracies $a_i$ above $a_{max}$ will always be rejected. Typical parameter values are

$v_{max} = 5km/h$ and $a_{max} = 50m$.