This rover definitely can fly in-game, but MechJeb shows that it has zero delta-V. There is a fair amount of clipping and weird attachment going on in there, but no cheats were used to create it (only possible editor bugs =)). It also uses that trick where a fuel line connected to a part makes it passable to fuel in general, which may be the cause of the problem. Interestingly, MechJeb 1.9.* does show that it has fuel.

http://filebin.ca/fNjUXtwHY4s

Btw, note how the mechjeb part has to be put on its side in order to provide a proper forward-looking control-from-here point; I've filed a bug about that a few weeks ago. With 1.9.* I used the other radial part here, but now it's gone.

I have noticed that the delta-V display goes to zero below a certain minimum threshold fuel level, guessing from https://github.com/MuMech/MechJeb2/blob/master/MechJeb2/FuelFlowSimulation.cs#L245 that it might be 1.00 units of fuel. Probably not what you're running into, but same symptom and has gotten annoying for me with some of my tiny probes recently.

@angavrilov, thanks for the test case; apparently I didn't properly understand the meaning of "fuelCrossFeed = false". The above commit fixes the issue for the posted craft. If you want you can grab the updated dev build .dll from http://jenkins.mumech.com/job/MechJeb2/

@tavert, yes the dV simulation treats any amount of fuel below a certain threshold as zero. In the above commit I lowered that threshold from 1 unit to 0.1 units, so you could also try out the dev build and see if it works better for you.