Informationsfusion für die kooperative Umfeldwahrnehmung vernetzter Fahrzeuge

Abstract

Cooperative decisions of multiple vehicles need a collaborative perception. Therefore, fusion methods for ego and environmental information are developed and evaluated. Precondition for the registration of data in a common coordinates is the accurate localization. This is realized by a fusion approach with absolute data of GPS and relative data of a dead reckoning system. A decentral fusion procedure combines a recursive multi-target tracking of detections with a grid-based mapping. Thus, negative sensor evidences can be formulated and missing detections are recognized.