This paper has presented a novel approach for detecting hazardous situations. The approach consists of assuming that hazards occur in atypical situations. For this reason, the normal space is defined and an atypical situation detector is developed. Since not all atypical events imply a danger, a further processing step deals with danger assessment by means of plausibility checks. In particular, this additional step helps to reduce the false alarm rate The system has been presented as a part of the multi-sensor multi-level fusion framework. An object detector has been presented to provide input information. It generates object hypotheses from the data of a laser scanner and a camera. To do this, the driving corridor is also detected. Finally, the hypotheses are fused by means ofan extended Kaiman filter Further, the information about the objects is projected onto an abstract layer, the Statistical Information Grid (SIG). If the SIG has been filled up with sufficient normal space data, it can be used for the detection of atypical situations during a drive. In future works, the detected atypical situations will be further processed so that dangers are assessed with more confidence. Besides, the presented SIG will be expanded to more information items coming from both surrounding objects and infrastructure information Furthermore, the expansion of the dimensionality ofthe SIG will be adhered In this first stage of the project, these prototypes are being developed and tested using the ViewCar, which gives a feeling of the kind of problems to deal with in real driving situations. However, the main aim of the work will be to analyse systematically the detection of actual dangers. For this future task, the driving simulator of the DLR ITS will be used, where dangerous situations can be artificially generated.
Early detection of hazards in driving situations through multi-sensor fusion
Frühe Erkennung von Risiken in Fahrsituationen durch Sensorfusion
2008
10 Seiten, 8 Bilder, 12 Quellen
Aufsatz (Konferenz)
Englisch
Early detection of hazards in driving situations through multi-sensor fusion
Kraftfahrwesen | 2008
|F2008-08-148 EARLY DETECTION OF HAZARDS IN DRIVING SITUATIONS THROUGH MULTI-SENSOR FUSION
British Library Conference Proceedings | 2008
|IDENTIFICATION OF HAZARDS IN URBAN DRIVING SITUATIONS USING ACCIDENT CATEGORIZATIONS
British Library Conference Proceedings | 2020
|Multi-sensor fusion for autonomous driving
TIBKAT | 2023
|