Efficient on-line path planning algorithm and navigation for a mobile robot

Abstract

A simple and efficient method for modelling a robot environment and an efficient on-line path planning algorithm are proposed to generate the optimal path within that environment. The suggested path planning algorithm is based on a good method for structuring the free space of a mobile robot environment into a set of overlapping prime rectangular areas PRAs that are suited to path planning and navigation tasks. The structuring of the free space environment starts by dividing the enlarged world map of that mobile robot environment into rectangular cells. These cells represent free and closed areas. Then, the free space in the specified environment is structured in terms of PRAs, and a connectivity table is constructed to provide the structure necessary for path finding. Using these results, a simple and computationally efficient algorithm is described which solves the pathfinding from a PRA containing the starting point to a PRA containing the goal point. Since movement between the PRAs cannot have fixed cost-distance attached to them, an efficient method is proposed that dynamically allocates costs to path segments as the path develops. The suggested algorithms allow optimal paths made up of straight lines and rectangular turns to be found efficiently with high speed. Consequently, a geometric representation of this path is executed. Then, a real time execution algorithm is introduced to control the robot during the execution of the planned path. Experimental examples based on the proposed algorithms and an analysis of the results are presented for a selected environment executed on the mobile robot Yamabico-9.