there is no way to actually get the coordinates in the simulation (how would the real robot know what position it is in?).
For the coordinates you could try to use color gradients in the background to determine the x and y position with the color sensor.
E.g. 0 - 255 red left to right, 0 - 255 green top to bottom, then use RGB mode of the colour sensor to determine the position. That way you would have different brightnesses of blue left for markers.
For the labyrinth in particular you could keep track of the turns already made and build up knowledge of already taken turns. This could be quite complicated though.