At its most basic, a cloaking device is some sort of system that diverts energy waves around itself without altering them, keeping whatever is inside the cloak from being affected by those energy waves. If we're talking about waves of light, that makes the contents of the cloak effectively invisible, since light encountering the cloak doesn't get absorbed by what it's protecting.
The best way that we've come up with to make something like this work is to use metamaterials, which is hard to do with light waves, because the structure of the metamaterial has to be a lot smaller than the wavelength of whatever you're trying to mess with to work properly, and visible light has very short wavelengths over a very long range.
There are plenty of other occasions where rendering something invisible to energy waves would come in handy, though, and seismic waves is a big one: with a seismic invisibility cloak, you have the potential to completely protect a structure from earthquakes, by using metamaterials to divert the seismic waves around an area. And since seismic waves have a much longer wavelength than light waves do, a seismic invisiblity cloak something that's actually not that hard to put together.
Researchers from the Institut Fresnel in Marseille and a French "ground improvement specialist company" called Menard have performed the first ever large-scale tests of a seismic cloaking device. The metamaterial cloak itself is deceptively simple: it consist of a carefully spaced array of five meter deep boreholes, and that's it. Here's a diagram:
Yep, as it turns out, a fancy deflector shield grid is completely unnecessary. You just need a bunch of empty holes. And it works. To test the cloak out, the researchers scattered an array of acoustic sensors all over their test area, and then started pounding the ground with something called a vibroprobe to simulate an earthquake. The metamaterial array of boreholes was able to redirect the seismic waves so well that acoustic sensors beyond the second row of boreholes were hardly able to detect anything at all.
There are just a couple problems with a setup like this: problem number one is that those redirected seismic waves have to go somewhere, meaning that your neighbors might get an extra nasty jolt during an earthquake thanks to the influence of your cloaking device. And problem number two is that cloaks work best when you know what the wavelength is that you're going to need to mess with, but earthquakes can come in lots of different wavelengths. A solution to problem one might be to fill those boreholes with something that will absorb seismic energy rather than reflect it, and a solution to problem two might be to tune the cloak to the resonance frequency of the structure that it's trying to protect.
Obviously, there's still work to be done here, but we know how to build seismic cloaking devices now, and most importantly, they're cheap and easy, so here's hoping that progress will be rapid. It seems likely that we'll see implementations of these cloaks within the next decade at least, probably protecting things like nuclear power plants and other high-value structures.