A Global Navigation Management Architecture Applied to Autonomous Robots in Urban Environments

This paper presents a global behavioral architecture used as a coordinator for the global navigation of an autonomous vehicle in an urban context including traffic laws and other features. As an extension to our previous work, the approach presented here focuses on how this manager uses perceived information (from low cost cameras and laser scanners) combined with digital road-map data to take decisions. This decision consists in retrieving the car's state regarding the global navigation goal, selecting which local navigation task should be executed (either lane following or intersection maneuvers), providing some round constraints and further defining the reference trajectory to be met by the selected local task. This system was experimented in a real autonomous car and provided satisfactory results.