Team 957 found success with our bridge manipulator as well -- but I wanted to share it because we did not use a wedge.
We had an arm that pivoted off a spot about two inches inside of the frame perimeter. It extended close to 14 inches out of the frame perimeter, and had the Tetrix wheels from last year's minibot kit. Therefore, I'd put the lever arm from the pivot to the contact point with the bridge at about 14 inches.
We originally planned to use a Banebots RS-550 with the AndyMark Gearbox and a Banebots 256:1 P80 (rated at 85 ft-lbs), but those were out of stock when it came time to order them. Instead, after doing the math and determining that this would be acceptable (if marginal), we used a kit Fisher-Price gearbox from a previous year (no gearing between the gearbox output and the arm's input). We strapped on a potentiometer for PID control of the arm.
We used a Jaguar to operate this motor, intending to use current control through CAN to limit our maximum current output. However, we wanted to have PWM control as a backup, and tuned that loop first. After finding out how fast that arm is*, as well as how noisy the data from the potentiometer is (I think we may have had a bad pot, too, which was replaced), I re-structured the code to allow limiting the commanded speed (not just duty cycle command) and to not rely on differentiating the potentiometer's output. We then relied on the Jaguar's dynamic braking to slow the arm down (and also relied on its linearity for our control loop tuning). This removed CAN-based current control from our plans. We limited our maximum duty cycle to 30% for testing with our bridge, but were prepared to raise it for competition.
Come competition, we went out on the practice field and tested with it a bit. At 50% duty cycle, it did not effectively push the bridge down (due to lack of torque). At 70%, the Jaguar's overcurrent protection kicked in well before the bridge even made it halfway down. At 60%, it was successful in operating the bridge (note: this will probably vary based on the motor used... I don't recall off the top of my head which we used). Although this warmed up the motor really quickly, we did not need to use the arm for more than a couple of seconds at a time, so it was not an issue during our matches.
If you do choose to use a Fisher Price gearbox-based arm, I'd recommend that you be ready to spend hours testing it, rewriting code, and tuning it. Also, verify (theoretically or practically) that the exact motor you use is capable of tilting a competition-weighted bridge.
A wedge is probably quicker and more effective at tipping the bridge, but may be harder to design (especially if, like us, only a few members of your team have the math knowledge to compute the required angles to tilt the bridge and handle the generated torque). Also, our arm is capable of lifting the bridge to let balls caught under it roll out. Had our drive team (which included me, our lead driver) known where the balls were getting stuck (right on the end, where they're in the way, invisible to the drive train, and not on the ramp), we could've swept them out from under the bridge, too.
* We have a video of the arm accelerating from a stop 5 inches above our bridge to an impact with our bridge in 3 frames... at 30 frames per second. The commanded rate limitation was implemented to decrease this impact. You can find this on YouTube at
http://www.youtube.com/watch?v=nXLTAL7Q3oo -- the fast motion is visible at about the 70-second mark.