Motion planning using a colored Kohonen network

Abstract

The motion planning problem asks for determining a collision-free path for a robot moving amidst a fixed set of obstacles. In this paper we present a new approach that combines a neural network and deterministic techniques to solve this problem. We define a colored version of a Kohonen network, which consists of two different classes of nodes. The network is presented random configurations of the robot and, from this information, it constructs a road map of possible motions in the work space. This road map can then be searched to find a motion connecting given source and goal configurations of the robot. The algorithm is simple and general; the only specific computation that is required is an intersection check for two polygons. It has been implemented for planar robots allowing both translation and rotation, and experiments show that compared to conventional techniques it performs well, even for difficult motion planning scenes. 1 Introduction The design of autonomous robots that are...