A motion plan is a multi-step process that can be used by a robot to precisely position itself to perform a specified task.
Suppose you own an autonomous mobile personal robot. You tell it to switch on the light in a bedroom. The light switch is on the wall. The robot has a computer map of the house stored in its controller memory. This three-dimensional map includes the exact coordinates of the bedroom light switch, and/or its position relative to other objects. The robot first proceeds to the general location of the switch. The program that facilitates this is known as a gross-motion plan.
In order to find the switch, and to position its end-effector (robot hand) in the right place to toggle it, a fine-motion plan is required. Machine vision can facilitate this. An example is the so-called eye-in-hand system, which allows the robot to recognize the shape of the switch and guide the end effector into place, down to millimeter displacements or smaller. Another method of executing the fine-motion plan involves tactile sensing, in which the robot's end effector gropes along the wall until it finds and throws the switch, just as you could find and throw it with your eyes closed once you were in its general vicinity.