What kind of sampling speed are you looking for? Based on the demo video on the link KJ6EAD supplied, that RFID solution didn't look like it had a very high sample rate (several seconds per scan is what it looked like, but I couldn't find any further detail so it's difficult to say for sure).
Have you thought about optical tracking - that is, set up multiple video cameras, perhaps at the upper corners pointing downward, tracking (3D triangulation) retroreflective dots on the rats? The only problem with this scheme is not being able to identify each individual rat. Perhaps active IR dots, sending out a pulse code which could be decoded as part of the video stream? Another method might be to have addressable active IR dots, tracked by the cameras, each lighted in-turn via an RF link. Occlusion of the dots (whether passive or active) could be an issue, but with enough cameras to provide overlap, and proper design of the testing area (ie - no tunnels or tubes allowed, for instance), most problems could be overcome.
Another possibility might be a magnetic system - the rats pack would house a small coil (consisting of three orthogonal coils) that could be pulsed via addressable RF means, like the active IR dots above, and pickup coils on the walls and ceiling would determine X/Y/Z position based on magnetic strength (at issue here would be the battery life of the man-packs, plus the noisiness inherent in the system that would need to be filtered out, plus the need to build a non-ferrous environment, among other problems).
If you needed heading or other orientation data, then an RF link back from the man-pack from an on-board IMU might work well with the above ideas. A great issue to overcome, no matter what system, would be battery life for the man-packs (but may lithium ion batteries would help there).
I don't think, though, that you're going to find much off-the-shelf hardware for this kind of tracking (with the exception of an all-optical system - those are sold, though mostly as a passive system - for motion and gesture capture systems - they aren't cheap, though). The RFID system will probably be the most ideal, depending on what the system is capable of in sample speed, and what kind of sample speed you need for your results.
If you come up with anything different, though - I would love to know about it; I've got an application in mind to develop that hinges on a similar need, and needs a very high sample rate (none of the above solutions would work for my need, unfortunately).