Recently two new technologies designed to measure vehicle speed have emerged, GPS and Inertial Navigation Systems (INS). In this article we take a look at the differences between velocity measurements made by GPS and those made by the RT3000 Inertial Navigation System.
The RT3000 has three angular rate sensors, or gyros, that track the Heading, Pitch and Roll angles of the vehicle. It has three accelerometers that measure the X, Y and Z accelerations of the vehicle. Using the angles a “virtual” level platform is computed so that the North, East and Down accelerations can be computed; these are then integrated to give velocity and position.
Integrating sensors leads to drift, so the RT3000 uses GPS to correct the drift. Using an extended Kalman filter the RT3000 is able to correct the drift in velocity and position. The Kalman filter is also able to correct the sensor errors and the Heading, Pitch and Roll angles.
Modern, survey-grade GPS receivers measure speed by tracking the carrier-phase of the signal from the satellite. Trees, buildings and bridges obstruct the direct path signal from the satellites, preventing the calculation of speed.
In these tests a high-quality, unfiltered, 20Hz survey grade GPS receiver was used to collect the GPS measurements.
Because the RT3000 uses its accelerometers to give velocity outputs, the velocity noise is very small. The accelerometers lead to a short-term velocity noise of about 1 mm/s over a 1 second period. Long-term drift is the dominant error that affects accuracy; the Kalman filter applies the GPS corrections smoothly so as to keep the short-term noise small.
The long-term accuracy of RT3000 velocity is the same as the long-term accuracy of GPS velocity. For the RT3100 system this is 0.1km/h and for the RT3002 system this is 0.05km/h.
The RT3000 does not have jumps or spikes associated with GPS only systems. It has a much higher bandwidth than GPS-only and far less short-term noise. On graphs it has a much smoother appearance.
The RT3000 is also able to accurately separate the forward and the lateral components of velocity, something that GPS-only systems cannot do.
In this article we have looked at five discrete situations and compared the RT3000 speed output with that of GPS-only.
In each of the tests we have looked at the results when the vehicle is freewheeling. This makes it easy to compare the two measurements. The tests have also been performed in winter, when tree cover has less effect on the GPS.
The RT3000 was mounted in the door pocket for convenience, a better solution would have been to use an RT-Strut, but this was not available. For accurate Pitch, Roll, Slip Angle and Acceleration measurements it should be rigidly mounted.
The afternoon when the tests were run was a good afternoon for GPS, with between 10 and 12 satellites visible in open-sky. This helps the GPS performance, but does little to the RT3100 system.
Open-Sky. In open-sky the GPS-only performs very well. There is very little noise in the GPS data and the two systems compare very well.
Single Tree. As the GPS passes a single tree (about 8s into the graph) it is possible to see an increase in the noise of the GPS. The RT3000 measurements remain noise-free.
Partial Cover. Partial cover can have different effects on GPS depending on how many satellites are affected. In summer the effect is greater since the leaves reflect the signal from the unobstructed GPS satellites.
Dense Cover. Some roads have a lot of cover, so much that GPS cannot make very many measurements, and the measurements made are poor
Bridge. Under bridges the GPS is lost for a moment. (This is a big bridge and no GPS stands a chance of making measurements in it). The RT3000 accurately measures the deceleration to acceleration without the GPS signal.