The proposed RRT-based planners are intended as global
planners, i.e., they should provide locomotion between any two
positions in an environment. The performance of the planners
for single start/goal positions has been presented in previous
experiments. However, this performance does not indicate average
performance, when multiple starts/goals are considered. For
example, let us consider an RRT-MP planner with only the ‘goahead’
primitive. This planner will be efficient in situations where
the goal configuration can be achieved by moving forward;
however, it cannot provide any solution when the goal is behind
the robot. To verify the ability of the RRT-MP planner to provide
global motion plans, an extensive test using multiple start/goals
needs to be performed.