I've created a new genome assembler named Elloreas (an abbreviation for ELongating LOng REad ASsembler). Simply speaking, it works in the following way:
1) You provide it with a starter. It can be a contig from some other assembly or just a random read.
2) It maps reads to the 3' end of the starter.
3) It calculates a consensus of reads that overhang the 3' end and extends the starter.
4) It goes to the step "2)".
There are many similar assemblers for short reads, like TASR, Mapsembler2, NovoPlasty and others. However, as far as I know, Elloreas is the first such assembler designed for long reads from PacBio and Nanopore.
A useful feature of Elloreas is that if it sees several alternative extensions for the contig, it will tell the user about their existence and give their sequences too.
You can read more about Elloreas here: https://github.com/shelkmike/Elloreas