Is it possible to project a magnetic field at a location in space?

This is possible, it's just very very difficult to do.

People regularly do this with the electric field of light to move dielectric particles (insulators) in the lab, and the technique is known as "optical tweezers":

http://en.wikipedia.org/wiki/Optical_tweezers

The reason you don't want to try this with magnetic particles is that the magnetic field of light is much weaker than the electric field, or rather that it doesn't interact very much with most magnetic materials.

Greg's answer above is half correct - The fact that these fields oscillate in time means that the applied force would oscillate as well, however, the field gradient is exploited instead to make optical tweezers work.

Sure, it's possible. That's how sets of cell phone radio towers work. In a typical setup, several of these towers are placed in a line, some distance outside of a town, such that the line is roughly perpendicular to the direction to the town. The placement of the towers is designed so that the signals reinforce each other as much as possible only in the town. That way, most of the cell towers' transmitted power gets directed toward the area where most people are actually using their cell phones, and you don't waste a whole lot of power by transmitting into open wilderness.

With the cell towers, though, the only things that really get pushed around by the EM waves are electrons in people's cell phones. Theoretically, there's no reason the same concept couldn't be used to move something larger, like a magnet, but you'd need vastly more powerful transmitters - perhaps more powerful than anyone knows how to build with current technology (depending on how large of a magnet you're trying to move, and how far away it is). Still, it's just an engineering problem.

Also, a side note: it's generally the electric field that's responsible for moving things around, not the magnetic field. Magnetic fields themselves don't transfer energy, they only change the direction of existing motion. But the two kinds of fields mix when you change your reference frame, so it's kind of a semi-arbitrary distinction anyway.

This can certainly be done, as induction chargers work pretty much that way. That said, this technique is not very efficient and it is hard to scale at large distances.

For large distances, one of the most popular ideas is using microwave transmission.

There's a very good article over at wikipedia with a round-up of all these technologies with pros and cons.