Is there an intuitive explanation for how loop closure reduces the landmarks of prior observations?
In particular the landmarks are connected to each other only through the observer, observing a previously seen landmark will reduce the variance of the robots location but how does this then propagate to the other landmarks observed in the past?
