As its name suggests, the path planner produces paths for the robot to follow. As such, it is used by those processes that move the robot from one position to another. The path planner takes as input a map of the environment produced by the mapper, a goal position, and an initial position (specified as triples of X, Y, and theta values). It returns a sequence of X, Y, theta triples that denote significant waypoints along the path generated by the planner.
Paths are generated in the following manner. A simple wavefront expansion[11] algorithm is used to generate a unimodal, potential field between the goal and initial position. Then, an initial path is generated by following the gradient of this field. Waypoints along the path are generated by starting at the initial position and selecting the last point on the path for which there exists an unobstructed, straight line path from the initial state to the goal. This point is marked as a waypoint and the process is repeated from the new waypoint until the goal position can be seen.