Future advanced driver assistance systems as well as autonomous vehicles are expected to further increase their areas of applicability. Reliable maneuver estimations are a prerequisite for many of the intended functionalities. Accordingly, maneuver estimation systems need to cover a wide range of scenarios. The majority of recently presented approaches are targeted at fixed scenarios. However, having specialized maneuver estimation systems covering each possible scenario is unrealistic. Therefore, we present an approach for tackling discriminative maneuver estimations in generic traffic scenes. It is based on reusable, partial classifiers that are combined online using a technique called pairwise probability coupling. As a result we are able to make discriminative maneuver estimations in generic traffic scenes. The benefits and applicability are presented on inner-city real-world data sets. Our evaluation indicates that the assembled probabilistic maneuver estimation is not only able to outperform generative models; it surpasses the performance of specially designed models due to the reduced complexities of the partial classifiers.
Probabilistic, discriminative maneuver estimation in generic traffic scenes using pairwise probability coupling
2016-11-01
207415 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
British Library Conference Proceedings | 2016
|Probabilistic maneuver recognition in traffic scenarios
TIBKAT | 2015
|Probabilistic Maneuver Recognition in Traffic Scenarios
GWLB - Gottfried Wilhelm Leibniz Bibliothek | 2014
|