A Fast Path-Planning Algorithm for a Mobile Robot Based on a Quadtree Representation

Abstract

Determination of a practical collision-free path for a mobile robot between start and goal positions in a workspace is central to design of an autonomous mobile robot. This paper presents a feasible path-planning algorithm which runs on the quadtree representation via a path graph. The quadtree representing the workspace is obtained from fats conversion of a real image taken through a camera on the ceiling. Thus, the algorithm can run even when obstacles in the workspace are shifted frequently. The quadtree also integrates obstacle regions and other regions in the workspace with a hierarchical structure in positioning. By using the hierarchical structure, the algorithm can select fast the practical collision-free path out of the quadtree without interference checking between the mobile robot and its obstacles. Moreover, the algorithm searches the path in the quadtree by way of the path graph. The path graph is initially defined by an arc with start and goal position's nodes, and the proposed algorithm gradually builds the path graph on the quadtree by the following processes synchronously: 1) Selection of the shortest path out of the path graph; and 2) Expansion of a part of the path graph along the shortest path so as to avoid the forbidden regions in the quadtree. By this synchronism, the algorithm can keep the size of the path graph as small as possible in connection with given two positions, and consequently it runs fast enough to be used practically even in any cluttered workspace. Finally, several experimental results show that the proposed algorithm is superior to some conventional algorithms in respect of the calculation time.