CODEWITHSUNDEEP
filtergpskalman-filter
malintha
malintha

Reputation: 176

Derive Kalman Process model from GPS data

This question might have addressed earlier, but I couldn't find one. I want to implement kalman filter to rectify GPS data. To implement kalman filter I need to create process model and measurement model. For the process model I derived this equation with use of motion equations s = ut + 1/2at^2 and v = u + at.

Process model 

[Lat(t)]         [1 1 0 0]  [Lat]                [ProcessNoise]
[d(Lat)(t)]  =   [0 1 0 0]  [d(Lat)]     +       [ProcessNoise]
[Lon(t)]         [0 0 1 1]  [Lon]                [ProcessNoise]
[d(Lon)(t)]      [0 0 0 1]  [d(Lon)]             [ProcessNoise]

Please note d(Lat) and d(Lon) is the rate of changing Latitude and Longitude respectively. And time period is 1. Since I'm not getting any control inputs that part of the kalman equation is dropped.

I'm reading a gpx file which has Lat, Lon, Velocity, Course, Ldop, Vdop, pDop values. I have 2 problems.

  1. Can I derive d(Lat) and d(Lon) with Course value and will they be precise?
  2. What values should I use in process noise matrix?

Upvotes: 1

Views: 452

Answers (1)

AlexWien
AlexWien

Reputation: 28727

No, all gps attributes are already heavily Kalman filtered (in the chip). Course and speed is not dependent on positional (lat, lon) change. speed is calculated by physical doppler shift effect. there are rumors that course calculation also uses doppler shift effects.

you might smooth the gps data, when applying an additional Kalman filter, but they will not get more acurate.

Upvotes: 1

Related Questions