Multi-target tracking algorithms using a probabilistic data association are state-of-the-art for current automotive safety applications. When using sensors with good resolution capabilities, probabilistic data association hypotheses are complete. However, with limited sensor resolution, measurements may not be resolved. This leads to degraded association hypotheses and wrong state innovations using standard algorithms. In this paper, the benefit of integrating sensor resolution in the data association and filter innovation is shown by a comparison of an advanced algorithm and the standard approach. The total root mean square error (RMS) of position, velocity, and time to collision (TTC) of all targets are decreased by the proposed algorithm when targets are merged. Thus, this paper suggests that sensor resolution should always be modeled by multi-target tracking algorithms in automotive environments to gain superior results.
Influence of sensor resolution on time critical automotive applications
Einfluß des Sensorauflösungsvermögen auf zeitkritischen Anwendungen im Fahrzeug
2010
6 Seiten, 10 Bilder, 2 Tabellen, 13 Quellen
(nicht paginiert)
Conference paper
Storage medium
English
Influence of sensor resolution on time critical automotive applications
Automotive engineering | 2010
|Programmable Aging Sensor for Automotive Safety-Critical Applications
Tema Archive | 2010
|High Resolution Piezo Film Sensor Systems for Automotive Applications
British Library Conference Proceedings | 2003
|Automotive sensor applications
Tema Archive | 1990
|