Autonomous navigation in urban areas using GIS-managed information

Abstract

This paper describes the usefulness of a Geographical Information System (GIS) for autonomous navigation of intelligent vehicles. In many urban applications the use of GPS alone is not sufficient and needs to be backed up with Dead-Reckoned (DR) sensors, map data and additional sensors like cameras or laser scanners. Geographical information can be used in two different ways. Firstly, preexisting features of the environment, such as roads, can be used as constraints in localisation space. Secondly, the geographical information can include landmark locations. The use of these two types of data is illustrated by a localisation system for urban areas: a laser scanner detects natural landmarks that are characterised during a learning phase. As the amount of data can be large, we propose a strategy for grouping the laser landmarks in enhanced local maps corresponding to the roads of a GIS layer, through the use of an L1 GPS receiver and DR sensors. Real experiments are reported to illustrate the performance of this approach.