Using Neural Gas Networks in Traffic Navigation

Abstract

The quality of navigation methods for mobile means depends first of all on description accuracy of their movement in a given area. This paper deals with Neural Gas (NG) networks whose role is creating of topologies for complex objects as for instance road networks of municipal communications where it is necessary to determine relations among individual elements (in our case a network of mutually interconnected communication nodes), too. The proposed approach combines besides NG networks also tree search algorithms, namely A*, hereby enabling to consider also various restraints, e.g. traffic rules, too. The performance of the proposed algorithm is shown on the road network of the city Košice in Slovakia.