[Cyris12]

I personally believe that sensors would be more useful if your target was moving or at an unknown location. As it stands, teams have the exact position of stationary objects. I'm guessing most people use sensors for two reasons:

1) To generate outputs that would have been too complicated or innacurate to write manually.
2) To have your robot adjust dynamically to changing situations.

As for number 1, the outputs generated aren't usually too complicated (ie the robot just has to move in a straight line). With a little testing, it's possible to determine these values. I personally prefer to have values I know work hard-coded into our program, as opposed to rolling the dice with sensors.

I believe number 2 is less of an issue than people think. I'm not sure that:
a) More than a few robots will be able to reach the center with enough force and time to disrupt you.
b) Sensors will help if you are moved. Most of the dead reckoning programs aren't terribly complicated. They don't tend to involve intricate movements that can be disrupted easily. If something can move your robot enough to seriously hamper the dead reckoning program, I'm not sure sensors would help.