NASA, MIT, and the ESA are working on something new, difficult, and extremely useful: satellites that swarm.
We're not talking about sets of multiple satellites which can all be coordinated from the ground to work together; that doesn't count, useful as it is, because it requires control from the ground. We're not even talking about the ESA's set of satellites named SWARM from 2004; they were another of the dozens of groups of ground-controlled devices launched over the last several decades (which include communications satellites, GPS satellites, etc.) We're talking about satellites that exhibit autonomous collective behavior: devices that don't have a whole lot of intelligence in each one, and don't need to communicate with each other or mission control to decide their actions, but when taken as a group they can complete complex tasks that are difficult or impossible for large single satellites (or astronauts) to achieve.
We've mentioned before the advantage of swarms: robustness, flexibility, and distribution. A swarm of bees can act like an organism the size of a brontosaurus, with hundreds of eyes, thousands of legs, and stingers galore; but a brontosaurus is too heavy to fly, too big to enter a honeycomb, and too metabolically expensive to survive a global climate disruption. A swarm of bees can travel fast, turn on a dime (in fact, can turn on several dimes and go off in completely separate directions), and can retain its functionality even if significant portions of it die. This last point is particularly valuable in space, where radiation concerns often relegate satellites to using ten-year-old computer technology for its robustness; wouldn't it be nice if you could use the latest, fastest CPU's and sensors instead, and just say "oh well" when 10% of them get fried by cosmic rays? For space applications, swarms have the additional advantage that tiny satellites can sometimes fit in the empty nooks and crannies left by other payloads being launched up, or crew quarters unused during takeoff; every cubic centimeter of launch costs money, just as every gram of weight costs money. Saving on both might provide such a cost advantage that swarms could become the preferred form of satellite decades from now; who knows?
So what's the state of the art in swarming satellites?
Not much, yet.
NASA and MIT's experiment is starting off modest: have a pint-sized satellite hang out inside the International Space Station, testing its ability to navigate autonomously. Then have a couple more, wired for group behavior like flying in formation. The devices (they're calling them "droids", a la Star Wars) are basically just platforms for thrusters, with software that can be changed to try different control algorithms. Then, when these basics are worked out, other devices could be built that would perform useful tasks. They imagine these mini-robots helping astronauts on spacewalks, or onboard space stations; they also imagine them building large self-assembled structures in orbit: "Software designed to control small satellites could just as well be used to maneuver the pieces of a spaceship together". Self-assembly might be the most useful function of swarming in space, because of the extreme difficulties involved in having astronauts working safely and productively on spacewalk, and because of the large sizes of desired structures compared to the limited space of launch vehicles.
The ESA doesn't have anything launched yet, but appears to be doing some great research into what they call distributed and autonomous control and navigation (the better to look serious and get grant money for, as opposed to MIT's Star Wars references, which I bet are what actually get the researchers excited). They speculate on the uses of swarming satellites:
"Large structures could be built, or many satellites could fly in formation to simulate the performance of much larger apertures than can be launched, whole, into space. Or the satellites might work in a coordinated way to explore many asteroids, providing a large cross section of comparable information rather than a snapshot of just one asteroid as happens with current missions."
The "performance of much larger apertures" is a reference to how a collection of small telescopes can mimic the performance of one huge telescope: the creatively-named Very Large Array in New Mexico combines 27 radio telescopes to become the equivalent of one telescope that has "the resolution of an antenna 36km (22 miles) across, with the sensitivity of a dish 130 meters (422 feet) in diameter."
In February the ESA's Advanced Concepts Team had a Workshop on Innovative Systems Concepts, which included swarming satellites and other topics that push the envelope of what's possible. I haven't been able to find published proceedings for the workshop, but here's a technical presentation by one of the groups describing the control algorithms & kinematics of formation flying. Presumably the same sorts of controls would also be useful for distributed exploration and self-assembly of large arrays or objects in space; just tweak the algorithms for different group behavior.
