Classification of sonar targets in air: A neural network approach

Patrick K. Kroh, Ralph Simon, Stefan J. Rupitsch

Research output: Contribution to JournalArticleAcademicpeer-review

Abstract

Ultrasonic sonar sensors are commonly used for contactless distance measurements in application areas such as automotive and mobile robotics. They can also be exploited to identify and classify sound-reflecting objects (targets), which may then be used as landmarks for navigation. In the presented work, sonar targets of different geometric shapes and sizes are classified with custom-engineered features. Artificial neural networks (ANNs) with multiple hidden layers are applied as classifiers and different features are tested as well as compared. We concentrate on features that are related to target strength estimates derived from pulse-compressed echoes. In doing so, one is able to distinguish different target geometries with a high rate of success and to perform tests with ANNs regarding their capabilities for size discrimination of targets with the same geometric shape. A comparison of achievable classifier performance with wideband and narrowband chirp excitation signals was conducted as well. The research indicates that our engineered features and excitation signals are suitable for the target classification task.

Original languageEnglish
Article number1176
Pages (from-to)1-17
Number of pages17
JournalSensors (Switzerland)
Volume19
Issue number5
DOIs
Publication statusPublished - 7 Mar 2019

Fingerprint

sonar
Sonar
Robotics
Ultrasonics
Classifiers
Air
Neural networks
Distance measurement
air
Research
Navigation
Acoustic waves
classifiers
Geometry
Sensors
landmarks
chirp
robotics
navigation
excitation

Bibliographical note

Special Issue: Eurosensors 2018 Selected Papers

Keywords

  • Feature extraction
  • Neural networks
  • Sonar detection
  • Sonar measurements

Cite this

Kroh, Patrick K. ; Simon, Ralph ; Rupitsch, Stefan J. / Classification of sonar targets in air : A neural network approach. In: Sensors (Switzerland). 2019 ; Vol. 19, No. 5. pp. 1-17.
@article{a807225f88ee434198e851cba3397a4d,
title = "Classification of sonar targets in air: A neural network approach",
abstract = "Ultrasonic sonar sensors are commonly used for contactless distance measurements in application areas such as automotive and mobile robotics. They can also be exploited to identify and classify sound-reflecting objects (targets), which may then be used as landmarks for navigation. In the presented work, sonar targets of different geometric shapes and sizes are classified with custom-engineered features. Artificial neural networks (ANNs) with multiple hidden layers are applied as classifiers and different features are tested as well as compared. We concentrate on features that are related to target strength estimates derived from pulse-compressed echoes. In doing so, one is able to distinguish different target geometries with a high rate of success and to perform tests with ANNs regarding their capabilities for size discrimination of targets with the same geometric shape. A comparison of achievable classifier performance with wideband and narrowband chirp excitation signals was conducted as well. The research indicates that our engineered features and excitation signals are suitable for the target classification task.",
keywords = "Feature extraction, Neural networks, Sonar detection, Sonar measurements",
author = "Kroh, {Patrick K.} and Ralph Simon and Rupitsch, {Stefan J.}",
note = "Special Issue: Eurosensors 2018 Selected Papers",
year = "2019",
month = "3",
day = "7",
doi = "10.3390/s19051176",
language = "English",
volume = "19",
pages = "1--17",
journal = "Sensors",
issn = "1424-8220",
publisher = "Multidisciplinary Digital Publishing Institute (MDPI)",
number = "5",

}

Classification of sonar targets in air : A neural network approach. / Kroh, Patrick K.; Simon, Ralph; Rupitsch, Stefan J.

In: Sensors (Switzerland), Vol. 19, No. 5, 1176, 07.03.2019, p. 1-17.

Research output: Contribution to JournalArticleAcademicpeer-review

TY - JOUR

T1 - Classification of sonar targets in air

T2 - A neural network approach

AU - Kroh, Patrick K.

AU - Simon, Ralph

AU - Rupitsch, Stefan J.

N1 - Special Issue: Eurosensors 2018 Selected Papers

PY - 2019/3/7

Y1 - 2019/3/7

N2 - Ultrasonic sonar sensors are commonly used for contactless distance measurements in application areas such as automotive and mobile robotics. They can also be exploited to identify and classify sound-reflecting objects (targets), which may then be used as landmarks for navigation. In the presented work, sonar targets of different geometric shapes and sizes are classified with custom-engineered features. Artificial neural networks (ANNs) with multiple hidden layers are applied as classifiers and different features are tested as well as compared. We concentrate on features that are related to target strength estimates derived from pulse-compressed echoes. In doing so, one is able to distinguish different target geometries with a high rate of success and to perform tests with ANNs regarding their capabilities for size discrimination of targets with the same geometric shape. A comparison of achievable classifier performance with wideband and narrowband chirp excitation signals was conducted as well. The research indicates that our engineered features and excitation signals are suitable for the target classification task.

AB - Ultrasonic sonar sensors are commonly used for contactless distance measurements in application areas such as automotive and mobile robotics. They can also be exploited to identify and classify sound-reflecting objects (targets), which may then be used as landmarks for navigation. In the presented work, sonar targets of different geometric shapes and sizes are classified with custom-engineered features. Artificial neural networks (ANNs) with multiple hidden layers are applied as classifiers and different features are tested as well as compared. We concentrate on features that are related to target strength estimates derived from pulse-compressed echoes. In doing so, one is able to distinguish different target geometries with a high rate of success and to perform tests with ANNs regarding their capabilities for size discrimination of targets with the same geometric shape. A comparison of achievable classifier performance with wideband and narrowband chirp excitation signals was conducted as well. The research indicates that our engineered features and excitation signals are suitable for the target classification task.

KW - Feature extraction

KW - Neural networks

KW - Sonar detection

KW - Sonar measurements

UR - http://www.scopus.com/inward/record.url?scp=85062833606&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85062833606&partnerID=8YFLogxK

U2 - 10.3390/s19051176

DO - 10.3390/s19051176

M3 - Article

VL - 19

SP - 1

EP - 17

JO - Sensors

JF - Sensors

SN - 1424-8220

IS - 5

M1 - 1176

ER -