Unlike RRT-MP, which is more efficient for a complex robot,
the success ratio of RRT-CPG decreases as the complexity of the robot increases. The position of the median indicates that for 50%
of the start/goal pairs, the success ratio is less than 75%. The success
ratio of RRT-CPG is higher for close start/goal pairs, but it decreases
with the distance between the start and the goal. This means that
the expansion step controlled by RRT-CPG provides only short
and inefficient local movements. RRT-CPG therefore needs more
iterations to grow the tree to the vicinity of the goal configuration.