I designed an extended kalman filter (EKF) for a multi-agent system which estimates a certain quantity which is common to all the agents of the formation (it is the scale of the formation, how big the formation is).
The EKF is working fine beside its initial behavior. Indeed as you can see from the picture below that at the beginning the estimation starts from ~5.5 and then it goes up to 11 while the signal is oscillating between 6.5 and 7.2.
Is it possible, without further details, to understand what is the problem in this Kalman Filter?
Thanks a lot for your help.
As pointed out in a comment there is no problem in my algorithm. It is just that the initial estimation which I assume in my EKF is too far from the actual state of my system.
What happens is that the EKF does not know where to go at the beginning and therefore it kind of explores a solution.