3

I'm interested in implementing a Kalman Filter in Python. First, I have programmed a very simple version of a K-Filter - only one state (Position in Y-Direction). My State transition Matrix looks like:

X <- X + v * t

with v and t are constants.

I simulate the measurement with a simple linear function

y = mx + b

and add noise to it:

y1 = np.random.normal(y, sigma, Nsamples).

It works pretty well, I can redefine R and Q to change the measurement and process noise value (Until now, it isn't a matrix).

Now I have an idea...

What happens if I would have a second measurement?

    y2 = np.random.normal(y, sigma2, Nsamples)

How could I handle it? Shall I prefilter measurement like :

(y1 + y2) / 2

or is there a more suitable method/solution which involves the kalman filter?

Mat
  • 63
  • 8

1 Answers1

1

There are numerous ways to handle fusion of multiple sensor measurements using Kalman Filter. A way to do it would be sequentially updating the Kalman Filter with new measurements.

See the slides by sensor fusion pioneer Hugh Durrant-Whyte found in this answer for quite a few ways how to fuse sensor data.

Malefitz
  • 417
  • 3
  • 7