Canu uses alignments between pacbio reads to correct away most sequencing errors in raw reads. You can't do much better without using the signal information. To achieve high-quality consensus, Quiver is your only choice.

The tools Pilon and Icorn2 performs assembly polishing with Illumina data. I am not sure if Icorn2 also takes PacBio reads for correction. Running additional rounds of Quiver may be the helpful to use the PacBio reads efficiently.