• We can assume that each primitive generates the desired
motion effectively. Such primitives hence move the robot
faster than randomly generated inputs. Planning with motion
primitives is thus faster and less memory consuming than naive
sampling-based planning with random control inputs.