Multisensorial data fusion for global vehicle and obstacles absolute positioning

Driving assistance is a popular research topic. Generally, two informations are necessary for a good driving: "where am I?", "where are dangerous obstacles on the road?". An approach which can precisely localize the vehicle and obstacles on the road is presented in this paper. Here, data fusion from proprioceptive and exteroceptive sensors like odometer, GPS, LIDAR, and vision as well as the knowledge of the road map allow this localization. This system is fully operational and has been tested in Clermont Ferrand city. Localization accuracy already reached, is a decimetric precision on lateral vehicle position and a metric precision on longitudinal vehicle position.