ABSTRACT
The obstacle avoidance scheme and the shortest path from the starting point to the finishing point is a common problem for autonomous Robot. This problem needs to be done in real time which becomes a challenge for large maps with complex environment as path planning is computationally very expensive [1]. One of the earliest and most well-known problems for such a system, especially for indoor domains, is the generation of collision free global path for the robot to move to a given point in a dynamic environment [2]. Conventional path planning approaches employed for such motion planning can be categorized into two types including global methods and local methods. The global methods such as road map [3], cell decomposition [4] and distance transform [5] which were able to search for possible paths in the whole workspace. The local methods have been proposed to provide eective path searching, such as the Potential Field method [6, 7] and the grid-based algorithms [8]. However, most of these algorithms suer from time ineciency in their computation and are not designed for use in real time path planning. Also, Potential Field methods are known to suer unwanted local minima in which the robot gets stuck in a U-shaped obstacle [9].
