From a computer/mathematical model, we have a very good idea of where the fuel mostly likely can and cannot be. The models run are not perfect, and there are uncertainties, but we know that units 2 and 3 couldn't have melted through the containment liner. It's very likely that unit 1 didn't melt through the containment liner either.
From an actual physical perspective, where exactly it is, you can't just go into containment. The dose rates and contamination levels are beyond acceptable for something like that. Additionally the fuel is not going to look like what you expect it to look like. We all need to remember that fuel < 5 years old, when it has little or no shielding, can deliver lethal doses in minutes.
TEPCO is looking into techniques to provide a definitive location of the core material.
http://www.japantimes.co.jp/news/20...ic-rays-peer-into-reactor-cores/#.U1rNOFVdWSo
This is one method they plan on using, which seems pretty cool, and has been tested. The issue is the cosmic rays and other stuff they are looking for have very low interaction rates, and as a result it will take months to collect a sufficient 'image' to 'develop' the picture.
I for one am glad TEPCO is not saying they "know" where the fuel is with certainty. There is a degree of confidence involved, but it's much better to prove it. This isn't like Chernobyl where you had all this piping and other stuff below the reactor where the fuel could potentially slump. The BWR containment system pretty much puts bounds on how far it could have migrated, as the further the fuel tries to migrate, the more concrete and other materials will mix in with it, which will effectively reduce the heat density and limit the maximum possible transit.