Goal-biased probabilistic foam method for robot path planning

Abstract

This paper presents an improved variation of Probabilistic Foam Method (PFM) for robot path planning. In PFM, a structure named probabilistic foam, formed by bubbles propagate through the free space from initial configuration to goal as a breadth-first search, obtaining a collision-free path. Although the method is able to obtain a navigable path, it is computationally expensive. We propose a new foam propagation approach inspired on random tree growth from RRT. Results from simulation experiments using 2D and 3D map show benefits with the new method.