Kalman filter--17. Gaussian motion

Looking back at the previous content, we know there is a measurement update and a Motion update (forecast).

Motion updates are done with full probability or an addition, and we've solved more complex situations.

I took it out and came up with the formula.

And it also implements this step in code.

The motion part doesn't want to go deep, it's a very simple step, let's write it down. Suppose you live in a world where the center point is your best estimate of where you are, and the Gaussian function is your error.

Suppose you move a certain distance to the right, and the motion itself contains its own error, then you arrive at a predictive position that adds the amount of exercise to the mean, and this adds an error to the initial error.

If you move the distance to the right (the Green line), you are expected to arrive at the place you want to be, but you will lose the information, because your movement would have lost information, showing the error here.

This is mathematically easy to achieve, the new mean equals the old mean plus action, often called U.

Suppose you exercise 10 meters here will be the Green line will be 10 meters, the new sigma squared equals the old Sigma Squared plus the motion of the Gaussian distribution variance.

That's all you need to know, it's just a summation.

Test:

Kalman filter--17. Gaussian motion