Based on the number of iterations, we can see that almost all
algorithms solve the problem in less that the allowed number of
iterations (k = 5000), which is also indicated by success ratios
close to 100%. The only exception is the RRT-K algorithm, which
fails to find a solution for the Quadropod robot, and the success
ratio is only 15% in this case. The RRT-MP algorithm outperforms
the other two methods in all measured aspects: it solves the problem
in the shortest time and it generates the fastest trajectories.
RRT-MP is able to find a solution for all robots in less than 20 iterations,
which indicates that its motion primitives are very efficient.
Although RRT-CPG can also provide a solution, its runtime is
significantly worse than the runtime of RRT-MP, especially for the
Lizard and Quadropod robots. While RRT-MP is able to move the
robot over long distances in each expansion step, the movements
achieved by RRT-CPG are less effective. This is indicated by the
highest number of iterations over all algorithms. The runtime and
the number of required planning iterations of RRT-CPG increase
with the size of the robot.