Astronaut Carl Walz restarted rail service aboard the international space station today, sending a $190 million flatcar creeping down the length of the lab's new S0 truss, from one station to another, after tests were interrupted earlier in the day by a subtle glitch.

The mobile transporter itself is working flawlessly, but the effects of weightlessness, which cannot be fully simulated on Earth, are interfering with the operation of sensitive position sensors that provide feedback on the cart's location. The problem only occurs during an automatic sequence to lock the mobile transporter down at a work site. Flight controllers are able to latch it in place by manually uplinking commands.

"All the electronics, all the systems, all the software on the mobile transporter are working very well. There are no glitches," said Ben Sellari, the S0 launch package manager. "This is not a software problem. The system is operating exactly as we instructed it to. I think what we're finding out as we go through this is how the mobile transporter works in zero G."

The transporter eventually will be used to carry the space station's Canadarm2 crane to 10 different work sites on the completed solar array truss. Two such work sites are available on the S0 truss element currently in place, one on each end of the 44-foot-long beam. The transporter is designed to latch itself down at each worksite with up to three tons of force to provide a stable platform for the crane during future operations when it will be used to move massive outboard truss elements and other components into place.

The transporter is equipped with two magnetic sensors that detect the presence of a pair of magnetized strips at each work site. When both strips are detected, the transporter's control software knows it is properly positioned to being the latch-down procedure. But during today's initial operation at work site 4 on the starboard side of the truss, the sensors apparently lost contact with the strips as the latches engaged, presumably because the cart floated upward very slightly.

After lengthy troubleshooting, engineers decided to order the latches to disengage and when they did, the correct position indicators reappeared. Confident the cart was in the proper position all along, controllers successfully sent commands to manually drive the latches closed to verify their operation. They then ordered the latches to disengage and Walz, sending commands through a laptop computer, ordered the transporter to move back down along the track to the far side of the truss to work site 5.

Again, the movement itself was smooth and again, the position indicators lost lock as the latches were engaged by the automatic software sequence. Ground controllers took over the operation as before, sending commands to manually engage the latches, before having Walz send the cart back to work site 4 where it will remain until the next shuttle visit.

"The mobile transporter, of course, is a very complex piece of machinery," Walz told . "I think it has some 20-odd motors that control its motion and also the latching systems. And when it did move, it was almost anticlimatic. It started to move very, very smoothly - and of course, very, very slowly - and then it got into position and started to latch and something went wrong with the automatic software. Bt the ground was able to latch it manually."

Asked if the initial failure made him a bit nervous, Walz said "we knew it wasn't going to keep rolling right off of S0. So we were fairly confident we were in a safe configuration. We called the ground and they started to do the troubleshooting. So our hearts really didn't skip a beat because we knew it wasn't going to continue moving. But we certainly were concerned, we always want to see the hardware check out the first time."

Sellari said engineers ultimately may decide to modify the automatic sequence software to ignore the temporary loss of data from the magnetic sensors during the latch-down procedure.