Brute Force Safecracking

If you wanted to get into a safe, but didn't know the combination, how would you crack it? The dumbest, but guarenteed to work, solution is to try every combination until one works. A human safecracker would get tired within a few hours of doing this, so Hack A Day reports someone automating this...with robots. The robot works faster than a human safecracker too.
The robot is a metal-and-plastic manipulator machine controlled by an embedded computer, and would fit in a backpack. If the thief is sneaky, and does this at a time when most people are asleep, and muffles the whirring noise made by the servos, he could sneak it into a bank at 2am in a backpack, muffle all noise in the area, let it grind away for 3 hours, grab the safe contents, grab the machine, and be gone by 6am. If he's stealthy enough, no one would even notice.
In some ways, I suppose this was inevitable. Cryptographic brute force is the only known way to solve NP based problems, and the only P based way to crack a safe would be to somehow figure out some pattern to the combination based on the manufacturer's serial number.