Feature cars will contain a wide range of driver assistance systems, which will improve the traffic safety by reducing the necessary driver's tasks and by support or warnings in dangerous situations respectively. A large amount of these systems will base on a detailed perception of the vehicle's environment. A complete environment perception system is introduced. It can serve multiple assistance systems simultaneously. This system is designed to handle the wide range of scenarios including intersections or dense traffic, which will occur during the usage of passenger cars. It uses a LIDAR sensor and a video camera as well as additional information of precise digital maps and data, which can be received by wireless communication. An object tracking is applied to the sensor data. Afterwards, several object features are extracted and used by a multi class object classification. The features contain information of the LIDAR sensor, the object tracking and digital maps. Additional features are obtained from image processing algorithms, which also benefit from the LIDAR data and can perform object detection with 3D object models. The classification algorithms combine statistical approaches and rule based a priori knowledge. Finally, additional information can be received by wireless communication and will be added to the sensor based environment representation. The complete system is applicable in real time. A detailed evaluation of all algorithms is given based on labelled real world test data.
Multisensorsystem zur erweiterten Fahrzeugumfelderfassung
2008
179 Seiten, Bilder, 164 Quellen
Theses
German
Multisensorsystem zur erweiterten Fahrzeugumfelderfassung
TIBKAT | 2008
|Fahrzeugumfelderfassung mit mehrzeiligen Laserscannern
Tema Archive | 2004
|Fahrzeugumfelderfassung mit mehrzeiligen Laserscannern
IuD Bahn | 2004
|PMD-basierte 3D-Optosensoren zur Fahrzeugumfelderfassung
Tema Archive | 2007
|