Path Planning for On-road Autonomous Driving with Concentrated Iterative Search
Qi Wang, Kuifeng Su, Dongpu Cao · 2019
This work proposes a novel method of generating an on-road path comprising simple, short line segments, guaranteeing both dynamic and obstacle constraints with enough smoothness. Instead of certain forms of spline, simple line segments are directly used to form the path. Initially, a very raw path is generated consisting of several rough line segments in sequence. The initial path represents a rough guess that lies close to the optimal solution with relatively high possibility. Therefore, subsequent search shrinks its sample space to the neighborhood of the generated path to produce a finer path comprising shorter line segments. The same process is repeated until the line segment is shorter enough and becomes negligible to be followed by the ego vehicle. The proposed method only uses simple line segments, but the result naturally forms a very smooth path that meets the constraints needed in the end.