[Al Skierkiewicz]

Evenstar,
I am surprised that you have had problems with batteries, especially the need to replace them every few runs during competition. Are you sure that your battery terminals were not deformed causing a high resistance connection to the batteries? Remember that the battery case is series wired and therefore there are twelve connection points to consider. The terminals are easily deformed with rough handling (mass vs acceleration) and dropping. If you turn the RCX over (without the battery cover in place) and batteries fall out, then the terminals are deformed. Damaged motors are another source of high current draw. Frequently a dropped motor will exhibit exceedingly high internal friction due to the rotor/frame spacing. A deformation here causes the rotor to rub on the frame. A simple test is to turn the motor shaft by hand. It should turn with a little resistance and continue turning when the shaft is released. If you cannot turn it by hand or it stops immediately when the shaft is released, then the motor is defective. As Danny's link above attests, these motors for the most part are not repairable. (I have had limited success realigning a bent motor mousing with the old motors.)
The Robolab battery meter is normally used with the robot idle (no motors running) so it is measuring no load battery voltage. It is still a good indication for the most part. A reading of less than 8 volts, no load, is a good indication that the batteries need replacing. Many FLL teams have not learned the direct relationship between large sticky tires and the high currents involved in turns. The large (3.5" diameter) sticky Lego tires on all four wheels with direct drive (wheels mounted directly on the motor shaft) is the highest current that can be drawn from the RCX. The addition of the friction imparted to the motor shaft by the direct tire mounting and the steering friction of turning sticky tires on the front of a robot combined with significant weight of robot design/attachments can significantly reduce battery life. As in FRC robots of old, high current draws during turns coupled with low battery can cause controller failure.
I have not heard of the need to change batteries during a competition other than to insure correct operation when the robot makes it to finals.