For moving robots or calculate routes to move from point A to point B, you need to use a graph data structure. A graph is a set of nodes and a set of edges that relate nodes. A path is a sequence of segments (i.e. two nodes and the edge used to go from the first to the second). A path finding algorithm then allows to compute an optimal route between 2 points.
If your robot moves on a flat and limited area (a floor, a building), a 2D map with X and Y coordinates for each node is sufficient. The nodes would represent key points for the possible routes (both ends of a corridor, center of doors, etc.. If needed a 3D structure could be used for representing the floor level.
If your robot moves on a larger scene, you then need to use GPS coordinates (latitude, longitude). From the graph algorithm point of view, it doesn't change much, it's just that the calculation of distance between points will get more complex.
In the latter case, you could be interested in using data formats in which you could easily get data, such as OpenStreetMap. It uses several data formats but one of it is OSM XML or its OSM JSON variant. But before reinventing the wheel, you may consider having a look at existing libraries/frameworks.
Finally, note that in real life the nodes are abstract points, and a road or corridor segment is larger than the theoretic line between two nodes. Here one possibility is to create a grid of tightly interrelated nodes. The other is to use the nodes to calculate the route, but let the robot deviate from the ideal theoretical trajectory, within some tolerances, based on the known topology and some sensors to avoid collisions.
But now we are entering in very complex aspects with a broad set of solutions.