Chair of
Multimedia Communications and Signal Processing
Prof. Dr.-Ing. André Kaup

EDAS: Error Detection & Archiving System

 

INI.FAU-Project EDAS: Error Detection & Archiving System

EDAS Image 

Project Description

Modern automobiles contain a variety of different components for both informing and entertaining the passengers. These components are managed by a central processing instance and are commonly referred to as infotainment systems in their entirety. Infotainment systems in cars provide a variety of services, ranging from mobile communication and media playback to television and parking assistance. The infotainment system presents all functions in a consistent user interface and renders all information in a visually appealing way. Of course, the growing complexity of interconnections between the different components as well as the rising number of the components comes with an increased probability of software and hardware errors.

Due to the necessity of error-free operation, the assessment of correct system behavior of these infotainment systems has high importance in the automobile industry. Typically, detailed integration tests on hard- and software are run and evaluated before vehicles are made available to the customer. Only after the systems have been extensively tested, a rollout to the car can take place. In order to span a wide variety of different test cases and devices during these tests, a fully automated integration and test system with emphasis on long term testing periods would provide a substantial benefit. Not only would such a system reduce the costs for monitoring and testing the systems, it would also allow to detect hard- and software problems in early development stages of these infotainment systems. It is essential for such an automated system for integration tests to allow recording of the data produced throughout the infotainment environment. Also, the setup must provide reliable algorithms for asserting error-free operation, i.e. provide tools to analyze the infotainment data in an automated way.

The INI-FAU project aims at providing such a fully automated integration and test system. With AUDI AG as a partner, the chair LMS designs, integrates and evaluates new algorithms for signal analysis and signal compression in the context of automotive error detection and archiving. For this purpose, a software module is developed, called Error Detection and Archiving System or short EDAS, which allows to record relevant video streams from the infotainment systems in a highly efficient, compressed form and use this archived data for running elaborate test and error detection algorithms in a fully automated test setup. Since the focus of the project covers both compression and error detection algorithms, it is subdivided into two subprojects, namely Compression of Display Data (Contact: Dominic Springer) and Automated Video Error Detection (Contact: Gilbert Yammine).

 

The EDAS Project structure comprising the two subprojects: Compression of Display

Data & Automated Video Error Detection

 



Compression of Display Data in AUDI Infotainment Systems
Contact: Dominic Springer

The goal of the compression subproject is to provide highly efficient compression capabilities for existing and future AUDI infotainment test setups. The developed compression algorithms also serve as a component to the Error Detection and Archiving System (EDAS), where they seamlessly integrate with the algorithms from the Automated Audio/Video Error Detection in order to provide fully automated error detection and archiving capabilities. The developed compression algorithms are designed specifically for AUDI infotainment test scenarios. High efficiency of the developed compression scheme is a central requirement and is addressed by a set of custom tailored compression schemes. An example for such a scheme, targeting the compression of navigation sequences from infotainment devices, can be seen below. By integrating higher order motion compensation modules into an HEVC codec, the required bitrate can be significantly reduced (click to enlarge):

    
Fig. 1: Example of a rotating frame pair from a navigation infotainment sequence (rotation clockwise). The rotation can not be compensated properly by standard video codecs and requires additional bitrate.  

Fig. 2: RDO results after encoding the left frame pair with standard HEVC. The encoder resorts to small-sized partitions and dominant intra modes, which are costly in terms of bitrate (red: intra, blue: inter, green: skip).

   

Fig. 3: Developed navigation sequence encoder for the EDAS project. The encoder is based on the HEVC architecture,  introduced components are highlighted in red.   Fig. 4: RDO results after encoding with the developed navigation sequence encoder. The encoder is able to compensate higher order motion (e.g. rotation), which results in efficient inter-predicted blocks and large partitioning.

The developed compression schemes are combined with the error detection algorithms (see below) to form the core of the EDAS software and provide a solution for a wide set of automotive test scenarios.


Automated Video Error Detection in AUDI Infotainment Systems
Contact: Gilbert Yammine

The goal of the error detection subproject is to provide testing algorithms that can be used in automated tests for the detection of possibly occurring errors on the main infotainment display of the car. The developed algorithms are embedded as single plugins in the error detection module of EDAS, which allows them to work in a parallel and automated way and makes the system more expandable. The possibly occurring errors are already known video errors that were observed once by testing engineers, but occur rarely and are difficult to reproduce. For that, tests should be run over long periods of time while the system should be automatically monitored for the occurrence of such errors. Because the infotainment system is a very complex unit providing different services for the user, and in order for the system to work in an automated way, the error detection module should first detect the context of the system (i.e. menu, video playback, navigation system...) in order to enable and/or disable the specific error detection plugins related to that context. Once an error is detected, it is logged by an error logger and the compression module is then signalled to start compression, when EDAS is run in online mode.

The architecture of the Error Detection Module of EDAS, which

is the main focus of the Automated Video Error Detection project


Publications:

2014-43
CRIS
D. Springer, C. Herglotz, F. Simmet, D. Niederkorn, A. Kaup
   [bib]

Real-Time Motion Classification of HD Video Sequences on Embedded Systems
IEEE European Embedded Design in Education and Research (EDERC), Pages: 157-161, Milano, Italy, Sep. 2014
2013-85 D. Springer, F. Simmet, D. Niederkorn, A. Kaup
   [bib]

Testgerät für ein Kraftwagennavigationsgerät und Verfahren zur Bewegungserkennung in einer Bildsequenz
DE102012023274, 2013
2013-84 D. Springer, F. Simmet, D. Niederkorn, A. Kaup
   [link]   [bib]

Method for Processing an Image Sequence and Tester for a Car
PCT/EP2012/004580, May 2013
2013-83 D. Springer, F. Simmet, D. Niederkorn, A. Kaup
   [link]   [bib]

Verfahren zum Verarbeiten einer Bildsequenz und Testgerät für einen Kraftwagen
DE102012009876, May 2013
2013-76
CRIS
D. Springer, M. Frank, F. Simmet, D. Niederkorn, A. Kaup
   [bib]

Spiral Search Based Fast Rotation Estimation for Efficient HEVC Compression of Navigation Video Sequences
Picture Coding Symposium (PCS), Pages: 201-204, San Jose, CA, USA, Dec. 2013
2013-44
CRIS
D. Springer, F. Simmet, D. Niederkorn, A. Kaup
   [bib]

Motion Vector Analysis Based Homography Estimation for Efficient HEVC Compression of 2D and 3D Navigation Video Sequences
IEEE Int. Conf. on Image Processing (ICIP), Melbourne, Australien, Oct. 2013
2013-43
CRIS
D. Springer, F. Simmet, D. Niederkorn, A. Kaup
   [bib]

Robust Rotational Motion Estimation for Efficient HEVC Compression of 2D and 3D Navigation Video Sequences
IEEE Int. Conf. on Acoustics, Speech and Signal Processing (ICASSP), Pages: 1379-1383, Vancouver, Kanada, May 2013
2013-22
CRIS
G. Yammine, E. Wige, F. Simmet, D. Niederkorn, A. Kaup
   [doi]   [bib]

A Novel Similarity-Invariant Line Descriptor for Geometric Map Registration
IEEE Int. Conf. on Image Processing (ICIP), Melbourne, Australia, Sep. 2013
2012-31
CRIS
G. Yammine, A. Khairat, F. Simmet, D. Niederkorn, A. Kaup
   [doi]   [bib]

Freeze Detection in 2D Navigation Video Sequences Overlaid with Real Satellite Images
IEEE Int. Workshop on Multimedia Signal Processing (MMSP), Pages: 43-48, Banff, Canada, Sep. 2012
2012-26
CRIS
G. Yammine, E. Wige, F. Simmet, D. Niederkorn, A. Kaup
   [doi]   [bib]

Freeze Detection in 2D Navigation Video Sequences by Matching of Extracted Line Segments
IEEE Int. Conf. on Vehicular Electronics and Safety, Pages: 55-60, Istanbul, Turkey, Jul. 2012
2012-23
CRIS
D. Springer, F. Simmet, D. Niederkorn, A. Kaup
   [bib]

Compression of 2D and 3D Navigation Video Sequences Using Skip Mode Masking of Static Areas
Picture Coding Symposium (PCS), Pages: 301 - 304 , Krakow, Poland, May 2012
2012-13
CRIS
G. Yammine, E. Wige, F. Simmet, D. Niederkorn, A. Kaup
   [doi]   [bib]

Blind Frame Freeze Detection in Coded Videos
Picture Coding Symposium (PCS), Pages: 341-344, Krakow, Poland, May 2012
2012-3
CRIS
D. Springer, F. Simmet, D. Niederkorn, A. Kaup
   [bib]

Compression of 2D Navigation Views with Rotational and Translational Motion
SPIE Electronic Imaging - Visual Information Processing and Communication III, Vol. 8305, San Francisco, CA, USA, Jan. 2012