The nearest neighbor rule used to select the nodes for expansion
ensures, that the RRT-MP can explore the configuration space. This
exploration is possible even if the utilized motion primitives move
the robot slowly. In such a case, the tree will grow slowly and it
will take more iterations to explore the configuration space and to
reach the goal configuration. Therefore, effective motion primitives
are preferred as they move a robot fast, which enables rapid growth
of the tree. The selection of the motion primitives is up to the user.
The primitives should be selected considering capabilities of the
robot as well as suitability of the primitives for the given task. For
example, primitives like ‘crawl’ or ‘turn’ are enough for moving on a
terrain, but ‘stand-up’ primitive may be necessary if the robot has
to reach a place above the ground. As the complexity of RRT-MP
is given by the amount of the motion primitives, the user should
select such a finite subset of primitives, that are believed to be
necessary to accomplish the task.