CODEWITHSUNDEEP
kalman-filter
nilkn
nilkn

Reputation: 985

Implementing a Kalman filter for position tracking given only position measurements (along with covariance)

I'm trying to track an object moving through space. The actual movement of the object should generally be fairly straight and even when not straight it should be smooth.

My measurements consist of the 3D coordinates of the object, the timestamp, as well as a 3x3 covariance matrix, but that's it. I do not have the velocity or acceleration (except insofar as it could be estimated from different position measurements).

Is it possible for me to use a Kalman filter with this data?

Upvotes: 4

Views: 1260

Answers (1)

dmuir
dmuir

Reputation: 4431

Yes.

I wouldn't bother faking up velocity observations as in effect the kalman filter will be doing that.

I'd guess you'd want position and velocity in the state vector; whether to have acceleration too is trickier; if the object is turning/accelerating slowly I'd first try not having acceleration in the state.

I've found that most of the work in implementing such filters goes into tuning, that is choosing, and perhaps adapting, the process-noise covariance matrix.

Upvotes: 4

Related Questions