Universality for unambiguous automata

Determine the complexity of universality problem for unambiguous finite automata. Problem is known to be in PTIME and (as far as I understand) in $\textup{NC}^2$, but is only known to be NL-hard.