By the way, NOVA doesn't seem to be maintained anymore, except a bit by Genode developers. Is this observation correct?
My observation is that most of the innovations were achieved by 2012 in microkernel research so stable work is what we see in Fiasco.oc and Nova. The only problem was that these OSes could not be formally verified. Muen answers the problem by hardcore simplicity, while I believe SEL4 has practicality issues.
Formal verification is not interesting for developers so somehow funds went somewhere else. Just a thought.
Rest is driver development which is not a research and showcase problem and handed by industry rather than universities,
TUD Genode people are the pioneers and can answer better.