The RRT-K approach is able to find solutions in the Steps and
Surface experiments, even if its success ratio is lower than the
success ratio of RRT-MP. However, as the number of modules of the
robots increases, the number of possible combinations of control
inputs grows exponentially. As RRT-K uses only a predefined fixed
number of K random combinations, it has less chance to generate
a good control input randomly. This slows down the exploration
process and consequently increases the time required to find a
solution.