I am not sure this can help, but I added some debugging output to r_c_shortest_paths.hpp. Here are my findings. The assertion fails when reconstructing pareto-optimal paths by walking back through pareto-optimal labels, at the end of the labelling process.

Walking back a pareto-optimal path: 4163 3567 Assertion failed: (p_cur_label->b_is_valid)

Ok, so what is the parent of this label 3567, that causes the assertion to fail?

New label 3567 is feasible and extends 3454

Ah-ah! And why is this label 3454 not valid?

Deleting dominated label: 3454 (dominated by 17903)

So, apparently a dominated label made it into a pareto-optimal path?!

Find attached r_c_shortest_pats.hpp modified to print debug info.