Found with ACube:
depth 18...
F2 U F' U' R U R' U' R U R' U' R U R' F U' . F2 (20q, 18f, 18s)
F2 U F' R U' R' U R U' R' U R U' R' U F U' . F2 (20q, 18f, 18s)
F2 U' R U' F U F' U' F U F' U' F U F' R' U . F2 (20q, 18f, 18s)
F2 U' R F U' F' U F U' F' U F U' F' U R' U . F2 (20q, 18f, 18s)
F' U' R F U F' R' U F2 R U R' F' U' R' F' R . (18q, 17f, 17s)
R' U F' R' U' R U' R' U2 R F U F' U2 F R . (18q, 16f, 16s)
R' F R U F R U' R' F2 U' R F U' F' R' U F . (18q, 17f, 17s)
R' F' U2 F U' F' R' U2 R U R' U R F U' R . (18q, 16f, 16s)
First one is the same as Robert's with shorter set-up. The 6th one is looking pretty good.
Try
R' U F' R' U' R U' R' U2 R y' R U R' U2 R B
or
x' U' Lw' U2 R U' R' y R' U2 R U R' U R F U' R
Has no one ever searched RUF E-perm by QTM on ACube before? I can't believe it. They should of done that a long time ago.
Edit:
By the way, does anyone use Nakajima's "bad" R perm?
http://www.youtube.com/watch?v=3XZPjdi6M_c&feature=channel_page
Edit 2:
R' U F' R' U' R U' R' U2 R y' R U R' U2 R B
Got 2.45 with this. It's about as fast as first alg in this thread, but my cube is getting really bad. the x'RUDu2 alg is still faster for me, but this one might get faster.
Edit 3:
F2 U F' U' R U R' U' R U R' U' R U R' F U' F2. Thanks Fanwuq! This one is rather nice
You could execute it like this:
(R2 U R' U') y (R U R' U' R U R' U' R U R' F) (U' F2)
That is much better.
Great. Now I know 7 E perms.
Edit4:
2.83 4th try. I'm definitely switching. The other algs were all at around 5 seconds on the 4th try.
2.20 24th try. Better than any alg except for the one that I've used for almost a year. And I can feel it get even better.
2.11