Lightsail (which uses regular physics) is doing that, launching a breadbox sail and a second proximity measuring cubesat along with it.
So there's the challenge of building the space-rated devices, with all the testing. It's got to have independent power, have a set of tests defining "it goes" and not, and you've got to consider that LEO isn't really space, it's freefall in a gravity well and magnetic field with tenuous atmosphere. And it's expensive!
That will all take a lot of bench testing. So the project needs access to shielded, vacuum rated facilities, and so on. Testing satellite engine concepts with a 2D model (effectively a huge air hockey table) is also common. This would all be done to both build the satellite and to prove the concept. So they need to do the bench work first.
Bit of a nitpick, but this is true of any object in "space", at least on sub-galactic scales. If it wasn't in freefall around Earth, it would be in freefall around the Sun, and if not that, the Milky Way.
You are correct about the magnetic field and magnetosphere however.
There is a venture, Cannae, that is attempting to launch a Cubesat with an incarnation of this device.
I'm not sure if they're funded, but the cost for a mission of this type can be only a few hundred thousand US$, so it's not an unreachable quantity by any means.
So there's the challenge of building the space-rated devices, with all the testing. It's got to have independent power, have a set of tests defining "it goes" and not, and you've got to consider that LEO isn't really space, it's freefall in a gravity well and magnetic field with tenuous atmosphere. And it's expensive!
That will all take a lot of bench testing. So the project needs access to shielded, vacuum rated facilities, and so on. Testing satellite engine concepts with a 2D model (effectively a huge air hockey table) is also common. This would all be done to both build the satellite and to prove the concept. So they need to do the bench work first.