Deep learning applied to scenario classification for lane-keep-assist systems

Halil Beglerovic, Thomas Schloemicher, Steffen Metzner, Martin Horn

Research output: Contribution to journalArticleResearchpeer-review

Abstract

Test, verification, and development activities of vehicles with ADAS (Advanced Driver Assistance Systems) and ADF (Automated Driving Functions) generate large amounts of measurement data. To efficiently evaluate and use this data, a generic understanding and classification of the relevant driving scenarios is necessary. Currently, such understanding is obtained by using heuristic algorithms or even by manual inspection of sensor signals. In this paper, we apply deep learning on sensor time series data to automatically extract relevant features for classification of driving scenarios relevant for a Lane-Keep-Assist System. We compare the performance of convolutional and recurrent neural networks and propose two classification models. The first one is an online model for scenario classification during driving. The second one is an offline model for post-processing, providing higher accuracy.

Original languageEnglish
Article number2590
JournalApplied Sciences
Volume8
Issue number12
DOIs
Publication statusPublished - 12 Dec 2018

Fingerprint

learning
Advanced driver assistance systems
Recurrent neural networks
sensors
Sensors
Heuristic algorithms
Time series
inspection
vehicles
Inspection
Deep learning
Processing

Keywords

  • Advanced driver assistance systems (ADAS)
  • Automated driving functions
  • Automated driving functions (ADF)
  • Deep learning
  • Scenario classification

ASJC Scopus subject areas

  • Materials Science(all)
  • Instrumentation
  • Engineering(all)
  • Process Chemistry and Technology
  • Computer Science Applications
  • Fluid Flow and Transfer Processes

Cite this

Deep learning applied to scenario classification for lane-keep-assist systems. / Beglerovic, Halil; Schloemicher, Thomas; Metzner, Steffen; Horn, Martin.

In: Applied Sciences, Vol. 8, No. 12, 2590, 12.12.2018.

Research output: Contribution to journalArticleResearchpeer-review

Beglerovic, Halil ; Schloemicher, Thomas ; Metzner, Steffen ; Horn, Martin. / Deep learning applied to scenario classification for lane-keep-assist systems. In: Applied Sciences. 2018 ; Vol. 8, No. 12.
@article{0e6b39744b034c56a19c97d715a9e0f5,
title = "Deep learning applied to scenario classification for lane-keep-assist systems",
abstract = "Test, verification, and development activities of vehicles with ADAS (Advanced Driver Assistance Systems) and ADF (Automated Driving Functions) generate large amounts of measurement data. To efficiently evaluate and use this data, a generic understanding and classification of the relevant driving scenarios is necessary. Currently, such understanding is obtained by using heuristic algorithms or even by manual inspection of sensor signals. In this paper, we apply deep learning on sensor time series data to automatically extract relevant features for classification of driving scenarios relevant for a Lane-Keep-Assist System. We compare the performance of convolutional and recurrent neural networks and propose two classification models. The first one is an online model for scenario classification during driving. The second one is an offline model for post-processing, providing higher accuracy.",
keywords = "Advanced driver assistance systems (ADAS), Automated driving functions, Automated driving functions (ADF), Deep learning, Scenario classification",
author = "Halil Beglerovic and Thomas Schloemicher and Steffen Metzner and Martin Horn",
year = "2018",
month = "12",
day = "12",
doi = "10.3390/app8122590",
language = "English",
volume = "8",
journal = "Applied Sciences",
issn = "2076-3417",
publisher = "MDPI AG",
number = "12",

}

TY - JOUR

T1 - Deep learning applied to scenario classification for lane-keep-assist systems

AU - Beglerovic, Halil

AU - Schloemicher, Thomas

AU - Metzner, Steffen

AU - Horn, Martin

PY - 2018/12/12

Y1 - 2018/12/12

N2 - Test, verification, and development activities of vehicles with ADAS (Advanced Driver Assistance Systems) and ADF (Automated Driving Functions) generate large amounts of measurement data. To efficiently evaluate and use this data, a generic understanding and classification of the relevant driving scenarios is necessary. Currently, such understanding is obtained by using heuristic algorithms or even by manual inspection of sensor signals. In this paper, we apply deep learning on sensor time series data to automatically extract relevant features for classification of driving scenarios relevant for a Lane-Keep-Assist System. We compare the performance of convolutional and recurrent neural networks and propose two classification models. The first one is an online model for scenario classification during driving. The second one is an offline model for post-processing, providing higher accuracy.

AB - Test, verification, and development activities of vehicles with ADAS (Advanced Driver Assistance Systems) and ADF (Automated Driving Functions) generate large amounts of measurement data. To efficiently evaluate and use this data, a generic understanding and classification of the relevant driving scenarios is necessary. Currently, such understanding is obtained by using heuristic algorithms or even by manual inspection of sensor signals. In this paper, we apply deep learning on sensor time series data to automatically extract relevant features for classification of driving scenarios relevant for a Lane-Keep-Assist System. We compare the performance of convolutional and recurrent neural networks and propose two classification models. The first one is an online model for scenario classification during driving. The second one is an offline model for post-processing, providing higher accuracy.

KW - Advanced driver assistance systems (ADAS)

KW - Automated driving functions

KW - Automated driving functions (ADF)

KW - Deep learning

KW - Scenario classification

UR - http://www.scopus.com/inward/record.url?scp=85058274646&partnerID=8YFLogxK

U2 - 10.3390/app8122590

DO - 10.3390/app8122590

M3 - Article

VL - 8

JO - Applied Sciences

JF - Applied Sciences

SN - 2076-3417

IS - 12

M1 - 2590

ER -