Comment We're being actively misled about purpose of this (Score 3, Insightful) 115
Satellites fail, for the most part, when their rechargeable batteries quit and/or their consumable manoeuvring fuel runs out. These are among the heavier components aboard many satellites, so our hypothetical 'repair and resupply' launch is already going to be costly and heavy before you add all that unique and highly flexible hypothetical manipulator hardware. From any sort of rational economic standpoint, if you're going to launch a heavy, expensive satellite, you might as well launch a replacement (with all-new hardware, up-to-date electronics, incorporating the lessons learned from the previous iteration, etc.) instead of trying to fix or cannibalize the dodgy one in orbit.
Trying to service multiple satellites with one launch of our Swiss-Army-knife repair droid gets even worse, because manoeuvring between orbits tends to be very costly in terms of fuel (prohibitively so if a significant change in inclination is contemplated) and therefore weight.
And how user-serviceable are most satellites? Anything that's already in space now (or that is likely to be launched in the next decade) hasn't been designed to be repaired, modified, or scavenged after launch. Are we really solving the 'space junk' problem if our repair droid is inadvertently leaving behind a cloud of dropped screws and broken hardware? One satellite is easy to track and avoid. A haze of screws and plastic chips is not--and will still put a hole right through the ISS.
The folks at DARPA are sometimes crazy, but they're not usually idiots. Presumably they've been able to come up with the same objections as Slashdotters, and they probably realized them faster than we did. So what's really going on?
1) A stripped-down version of this tool could be used to attach de-orbiting or manoeuvring thrusters to disabled satellites that happened to be occupying (or threatening) particularly high-value orbital real estate. The ISS has to be periodically repositioned to avoid the occasional bit of space junk. Further up, there's a limited amount of space in geostationary orbit, and a malfunctioning satellite could be trouble as either a source of physical or radio clutter. If the program fails to produce its rather pie-in-the-sky 'dream' goal, it could still develop this useful sideline.
2) The military would love to have the capability to selectively damage, disable, and/or capture 'enemy' space hardware. This program would complete nearly all the steps required to develop such a capability, but under the shiny, happy patina of putative civilian applications.