(An) efficient path finding algorithm based on raster representation and visibility polygon

Abstract

In this thesis we develop an efficient path planning algorithm for mobile robot in the two dimensional space. Using the visibility polygon concept and raster representation of the objective world, we can reduce computational cost in two respects, expansion of only necessary portion of the search graph and efficient exploration of a node. Among child nodes of a node some nodes can not be included in the shortest path. In the proposed algorithm are generated only the portion of child nodes that have possibility to be included in the shortest path. Due to this restricted generation of child nodes, considerable reduction of the search graph is achieved. In spite of intrinsic error, raster representation is useful for path planning and has two advantages : fast determination of visibility of two points and no restriction on the shape of obstacle. Utilizing the first we can construct a visibility polygon from a point very efficiently. This means fast generation of child nodes. By the second advantage we can handle complicated environment with small effort.