Identification of Irregular Motion in Automotive Navigation Systems Using Novelty Detection

Pöllot M, Springer D, Schleifer R, Niederkorn D, Kaup A (2016)


Publication Language: English

Publication Type: Conference contribution, Conference Contribution

Publication year: 2016

Pages Range: 1-8

Event location: Athens GR

ISBN: 978-1-5090-4240-1

DOI: 10.1109/SSCI.2016.7850000

Abstract

Automated display testing for visual unpleasant and erroneous navigation sequences is an important step to preserve a high quality standard for premium vehicle manufacturers. This paper presents a novel error detection algorithm for navigation sequences based on novelty detection on motion parameters obtained from real world navigation sequences. Motion parameters are accumulated through key point matching using BRISK and subsequent homography calculation. With these parameters one is able to describe the motion between two successive frames. Combinations of translational and rotational components allow the novelty detection algorithm to predict outliers. These outliers either show positioning errors or abnormal motion behaviour which are both unacceptable for high quality. Experimental results demonstrate that this algorithm works significantly better than state of the art, where one has to know errors before analysing the data set in order to determine thresholds for particular errors. The gains in precision and recall are 49.67% and 6.06% respectively, the accuracy is 1.37% higher compared to optimized threshold results.

Authors with CRIS profile

Involved external institutions

How to cite

APA:

Pöllot, M., Springer, D., Schleifer, R., Niederkorn, D., & Kaup, A. (2016). Identification of Irregular Motion in Automotive Navigation Systems Using Novelty Detection. In IEEE (Eds.), Proceedings of the IEEE Symposium Series on Computational Intelligence SSCI (pp. 1-8). Athens, GR.

MLA:

Pöllot, Martin, et al. "Identification of Irregular Motion in Automotive Navigation Systems Using Novelty Detection." Proceedings of the IEEE Symposium Series on Computational Intelligence SSCI, Athens Ed. IEEE, 2016. 1-8.

BibTeX: Download