Hello,

I am studying graph-based SLAM from several papers, I struggle in understanding the concept of a virtual measurement zjk between two nodes xj and xk.

I try to explain: SLAM (simultaneous location and mapping) is a technique in robot vision so that a robot can simultaneously estimate a map of the world around and its position within this map.

In the graph-based formalism, a graph is constructed in which each node represents a position xk of the robot. To that position, a measurement zk of the world is made, and a probability p(zk | xk, m) is assumed to be known, where m is the map. Fair enough until here.

However, at a certain point, the concept of "virtual measurement" zjk between nodes xj and xk is introduced, and it is associated to an edge in the graph between nodes j and k.

Where does that virtual measurement come from? Does it relate to measurements zj and zk? how?

I thank you in advance.

Giuseppe