It's been quite some time since this post so generally I wouldn't advocate bringing it back from the dead but it was never responded to and I have some additional information to provide.
I was in fact able to do this however the max distance before seeing the wall sufficiently was about six inches. However, it's important to note that the wall was only one inch tall. I didn't test this method with a taller wall but I would assume it would be slightly better since you would obviously get more reflectance.
Another thing I learned from this is that using high wattage LEDs may not be feasible depending on your robot's size, it consumes a lot of power in addition to requiring various 3W resistors which are large in and of themselves. Ultimately it was a success with a robot about 9x6" in size.
I had two 1W LEDs facing outward at about 45 degree angles and below them I had five light intensity sensors (TSL235R) and I implemented a PID algorithm and I had tweaked the Pololu QTR library to work with the TSL sensors so essentially it was almost exactly like using a pololu line sensor (from a software perspective).
I had a fun time getting this to work but I feel there are better options to go with given the power consumption and minimal distance to the short wall. For me this was merely a proof of concept so I didn't take it much further and I didn't look into alternatives so I can't give much in the way of suggestions there.
The suggestions I can provide are that you limit the angle at which light comes into the sensors so you know only mostly reflected light will be coming in. Keep in mind the distance at which you need reflectance (greater than 6" is difficult with a short wall). Maybe use a lens on the LED to focus the beam, I bought some but never had time to try them out, I feel this would give better range since you can focus the beam to get higher reflectance.
That's all for now, good luck and I'll try to keep checking this but you may want to start your own thread once you get started.