This is a general logic question.

I have this robot for use around the office. Our office room is a sort of maze with the layout of the cubicles and such. Everyone in the office has a program that they can use to call the robot to their cubicle.

What I am trying to figure out is the most logical way to give the robot a little AI in knowing where he is located in the room, and the best possible way to get to the next destination from there. As it stands right now it has a preprogrammed paths, so it can only travel from certain destinations, to certain other ones. It uses a system of exact measurements and sensors to avoid obstacles. This poses a problem if the robot is not placed exactly right, if it has to stop if someone walks near it, and sometimes it just gets confused and goes wacky...

any suggestions? I'm thinking of making a map with plot points. but i don't know where to go from there

Map!!..I guest that will be hard. Using sensor hack backlog..it is indeed the consiquence of it. Are you building a surveilance mobo?!.

This is a very difficult problem whose solution will not fit into messages on a forum like this one. Some important details to consider are: 1. Sensor modalities and quality available to the robot, 2. On-board processing power and memory available and 3. Locomotion capabilities of the robot. As an off-the-cuff (!) suggestion, look into occupancy grids.