I am currently working with kalman filter in area of target tracking. The sensor we are using gets me multiple measurements at each time step. Number of measurements is not fixed, meaning that in one step I could get N Points, while in the next step I could get any other number of Points M.
Now I have a working implementation of KF and I am using sequential update procedure to update the filter with multiple measurements. What I mean is that I perform multiple update steps ( one per measurement ) before doing the next prediction step.
Now I am looking to transition into adaptive KF and for that I need the smoother, and smoother has the same problem and honestly I have not really troubled with the smoother so far. Before starting the heavy Lifting I would like to learn some general methods of integrating multiple measurements into these filters at each time step?
I am assuming complete re-derivation of KF would be necessary, which I am comfortable with. Literature is welcome as well.
Thanks in advance.