Path planning for mobile robots based on a modified potential model

Abstract

The artificial potential field (APF) method is widely used for mobile robots path planning due to its elegant mathematical analysis and simplicity. However, this method has some inherent disadvantages in path planning. In this paper, a modified potential model is proposed to overcome these disadvantages. The new attractive function decreases the potential around goals evidently to eliminate the problem of goals nonreachable with obstacles nearby (GNRON). And a new repulsive potential field is established as discretization of the obstacles with points distributed over their geometry. Therefore, the environment is described more exactly by the new model composed of the new attractive and repulsive potential field. The simulation has proved that the new potential model can solve the GNRON problem, U-trap, and oscillations. In dynamic environment, the validity of the new potential model is also verified by simulation.