Path planning in a 2-D known space using neural networks and skeletonization

Abstract

A neural network and a skeletonization based path planning in a 2D known space is presented. For the neural network path planning approach a Kohonen self-organizing net has been chosen, while for the skeletonization Kwok's method (1988) was used. The output of the network represents a reduced representation of the free space available for robotic movement in a 2D known environment.