roadmap with curvature larger than ÖÑ Ò . After that, we
use Dijkstra’s algorithm to find the shortest path between
the start and the goal. The path will consist of a sequence of
nodes from the roadmap. Note that such a path could pos-
sibly contain several cusps, which indicate transitions be-
tween forward and backward movements of the car. From
the perspective of the control roadmap, the path is deter-
mined by a sequence of control points. After we have a
path, we must validate it with finer resolution collision tests
since the validity of the edges has not been verified before.