|
|
|
Here we will discuss the PVT calculation in dynamic mode with GPS/GALILEO/BEIDOU. This calculation is more complicated than its static counterpart because the element to be geolocated is in motion.
|
|
|
|
|
|
|
|
**A different state vector:**
|
|
|
|
|
|
|
|
Because of the motion, we have to incorporate new parameters in the state vector:
|
|
|
|
|
|
|
|
State vector 
|
|
|
|
|
|
|
|
With:
|
|
|
|
|
|
|
|
-  correspond to the speed of the receiver
|
|
|
|
|
|
|
|
- correspond to the coordinates of the receiver
|
|
|
|
|
|
|
|
-  are the bias and clock drift
|
|
|
|
|
|
|
|
-  correspond to the time lag between the different GNSS
|
|
|
|
|
|
|
|
**The steps of the dynamic position calculation :**
|
|
|
|
|
|
|
|
Thanks to this state vector, we can initialise the Kalman filter calculation process. Here is the order of the steps:
|
|
|
|
|
|
|
|
- The initialization phase with the state vector and the covariance matrix (see more details in the Kalman section)
|
|
|
|
|
|
|
|
- Prediction phase: the equations of motion of the object are used to predict the next position, based on the current position and velocity.
|
|
|
|
|
|
|
|
- Measurement: The current position of the object is then measured using a sensor
|
|
|
|
|
|
|
|
- Update: Finally, the Kalman formula is used to update the position estimate by combining the information from the prediction and the measurement.
|
|
|
|
|
|
|
|
Note that this process is repeated with each new sensor measurement. We can then position a moving element.
|
|
|
|
|
|
|
|
**Limitations:**
|
|
|
|
|
|
|
|
However, for an optimal estimation of the positioning, a good modelling of the system and the measurement noise is required. A bad modelling of these two elements can lead to more or less accuracy errors.
|
|
|
|
|
|
|
|
There will always be a small accuracy error, even with the most powerful applications. Let show this through an experience.
|
|
|
|
|
|
|
|
**Experience :**
|
|
|
|
|
|
|
|
The experiment conducted by the GEOLOC laboratory of the Gustave Eiffel University shows this.
|
|
|
|
|
|
|
|
On a non-linear path of 6.5 minutes in dynamic mode, in dual frequency, at a speed of 1.2 m/s, the Kalman calculation of the individual's geolocation gave different errors according to the different software used:
|
|
|
|
|
|
|
|
|
|
|
|
| Application | Average error in meter|
|
|
|
|
|--|--|
|
|
|
|
| GeolocPVT | 2.16|
|
|
|
|
| GalileoPVT|25.84
|
|
|
|
| GNSS Compare| 17.51 |
|
|
|
|
|
|
|
|
|
|
|
|
---- Rajouter l'image quand on fera la mise en forme ---- |
|
|
|
\ No newline at end of file |
