VANET meets deep learning: the effect of packet loss on the object detection performance

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

Abstract

The integration of machine learning and inter- vehicle communications enables various active safety measures in internet-of-vehicles. Specifically, the environmental perception is processed by the deep learning module from vehicular sensor data, and the extended perception range is achieved by exchanging traffic-related information through inter-vehicle communications. Under such condition, the intelligent vehicles can not only percept the surrounding environment from self-collected sensor data, but also expand their perception range through the information sharing mechanism of Vehicular Ad-hoc Network (VANET). However, the dynamic urban environment in VANET leads to a number of issues, such as the effect of packet loss on the real-time perception accuracy of the received sensor data. In this work, we propose a point cloud object detection module via an end-to-end deep learning system and enable wireless communications between vehicles to enhance driving safety and facilitate real-time 3D mapping construction. Besides, we build a semi- realistic traffic scenario based on the Mong Kok district in Hong Kong to analyze the network performance of data dissemination under the dynamic environment. Finally, we evaluate the impact of data loss on the deep-learning-based object detection performance. Our results indicate that data loss beyond 50% (which is a common scene based on our simulation) can lead to a rapid decline of the object detection accuracy.

LanguageEnglish
Title of host publication2019 IEEE 89th Vehicular Technology Conference, VTC Spring 2019 - Proceedings
Place of PublicationPiscataway
PublisherInstitute of Electrical and Electronics Engineers
Number of pages5
ISBN (Electronic)978-1-7281-1217-6
DOIs
StatePublished - 1 Apr 2019
Event89th IEEE Vehicular Technology Conference, VTC Spring 2019 - Kuala Lumpur, Malaysia
Duration: 28 Apr 20191 May 2019

Conference

Conference89th IEEE Vehicular Technology Conference, VTC Spring 2019
Abbreviated title VTC Spring 2019
CountryMalaysia
CityKuala Lumpur
Period28/04/191/05/19

Fingerprint

Vehicular ad hoc networks
Vehicular Ad Hoc Networks
Object Detection
Packet Loss
Packet loss
Learning systems
Communication
Sensors
Sensor
Intelligent vehicle highway systems
Safety
Traffic
Real-time
Intelligent Vehicle
Data Dissemination
Network performance
Module
Information Sharing
Point Cloud
Network Performance

Keywords

  • 3D Point Cloud
  • Autonomous driving
  • Deep learning
  • SUMO
  • VANET

Cite this

Wang, Y., Menkovski, V., Ho, I. W. H., & Pechenizkiy, M. (2019). VANET meets deep learning: the effect of packet loss on the object detection performance. In 2019 IEEE 89th Vehicular Technology Conference, VTC Spring 2019 - Proceedings [8746657] Piscataway: Institute of Electrical and Electronics Engineers. DOI: 10.1109/VTCSpring.2019.8746657
Wang, Yuhao ; Menkovski, Vlado ; Ho, Ivan Wang Hei ; Pechenizkiy, Mykola. / VANET meets deep learning : the effect of packet loss on the object detection performance. 2019 IEEE 89th Vehicular Technology Conference, VTC Spring 2019 - Proceedings. Piscataway : Institute of Electrical and Electronics Engineers, 2019.
@inproceedings{f0934f738e564c2ca4e618d20fc90631,
title = "VANET meets deep learning: the effect of packet loss on the object detection performance",
abstract = "The integration of machine learning and inter- vehicle communications enables various active safety measures in internet-of-vehicles. Specifically, the environmental perception is processed by the deep learning module from vehicular sensor data, and the extended perception range is achieved by exchanging traffic-related information through inter-vehicle communications. Under such condition, the intelligent vehicles can not only percept the surrounding environment from self-collected sensor data, but also expand their perception range through the information sharing mechanism of Vehicular Ad-hoc Network (VANET). However, the dynamic urban environment in VANET leads to a number of issues, such as the effect of packet loss on the real-time perception accuracy of the received sensor data. In this work, we propose a point cloud object detection module via an end-to-end deep learning system and enable wireless communications between vehicles to enhance driving safety and facilitate real-time 3D mapping construction. Besides, we build a semi- realistic traffic scenario based on the Mong Kok district in Hong Kong to analyze the network performance of data dissemination under the dynamic environment. Finally, we evaluate the impact of data loss on the deep-learning-based object detection performance. Our results indicate that data loss beyond 50{\%} (which is a common scene based on our simulation) can lead to a rapid decline of the object detection accuracy.",
keywords = "3D Point Cloud, Autonomous driving, Deep learning, SUMO, VANET",
author = "Yuhao Wang and Vlado Menkovski and Ho, {Ivan Wang Hei} and Mykola Pechenizkiy",
year = "2019",
month = "4",
day = "1",
doi = "10.1109/VTCSpring.2019.8746657",
language = "English",
booktitle = "2019 IEEE 89th Vehicular Technology Conference, VTC Spring 2019 - Proceedings",
publisher = "Institute of Electrical and Electronics Engineers",
address = "United States",

}

Wang, Y, Menkovski, V, Ho, IWH & Pechenizkiy, M 2019, VANET meets deep learning: the effect of packet loss on the object detection performance. in 2019 IEEE 89th Vehicular Technology Conference, VTC Spring 2019 - Proceedings., 8746657, Institute of Electrical and Electronics Engineers, Piscataway, 89th IEEE Vehicular Technology Conference, VTC Spring 2019, Kuala Lumpur, Malaysia, 28/04/19. DOI: 10.1109/VTCSpring.2019.8746657

VANET meets deep learning : the effect of packet loss on the object detection performance. / Wang, Yuhao; Menkovski, Vlado; Ho, Ivan Wang Hei; Pechenizkiy, Mykola.

2019 IEEE 89th Vehicular Technology Conference, VTC Spring 2019 - Proceedings. Piscataway : Institute of Electrical and Electronics Engineers, 2019. 8746657.

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

TY - GEN

T1 - VANET meets deep learning

T2 - the effect of packet loss on the object detection performance

AU - Wang,Yuhao

AU - Menkovski,Vlado

AU - Ho,Ivan Wang Hei

AU - Pechenizkiy,Mykola

PY - 2019/4/1

Y1 - 2019/4/1

N2 - The integration of machine learning and inter- vehicle communications enables various active safety measures in internet-of-vehicles. Specifically, the environmental perception is processed by the deep learning module from vehicular sensor data, and the extended perception range is achieved by exchanging traffic-related information through inter-vehicle communications. Under such condition, the intelligent vehicles can not only percept the surrounding environment from self-collected sensor data, but also expand their perception range through the information sharing mechanism of Vehicular Ad-hoc Network (VANET). However, the dynamic urban environment in VANET leads to a number of issues, such as the effect of packet loss on the real-time perception accuracy of the received sensor data. In this work, we propose a point cloud object detection module via an end-to-end deep learning system and enable wireless communications between vehicles to enhance driving safety and facilitate real-time 3D mapping construction. Besides, we build a semi- realistic traffic scenario based on the Mong Kok district in Hong Kong to analyze the network performance of data dissemination under the dynamic environment. Finally, we evaluate the impact of data loss on the deep-learning-based object detection performance. Our results indicate that data loss beyond 50% (which is a common scene based on our simulation) can lead to a rapid decline of the object detection accuracy.

AB - The integration of machine learning and inter- vehicle communications enables various active safety measures in internet-of-vehicles. Specifically, the environmental perception is processed by the deep learning module from vehicular sensor data, and the extended perception range is achieved by exchanging traffic-related information through inter-vehicle communications. Under such condition, the intelligent vehicles can not only percept the surrounding environment from self-collected sensor data, but also expand their perception range through the information sharing mechanism of Vehicular Ad-hoc Network (VANET). However, the dynamic urban environment in VANET leads to a number of issues, such as the effect of packet loss on the real-time perception accuracy of the received sensor data. In this work, we propose a point cloud object detection module via an end-to-end deep learning system and enable wireless communications between vehicles to enhance driving safety and facilitate real-time 3D mapping construction. Besides, we build a semi- realistic traffic scenario based on the Mong Kok district in Hong Kong to analyze the network performance of data dissemination under the dynamic environment. Finally, we evaluate the impact of data loss on the deep-learning-based object detection performance. Our results indicate that data loss beyond 50% (which is a common scene based on our simulation) can lead to a rapid decline of the object detection accuracy.

KW - 3D Point Cloud

KW - Autonomous driving

KW - Deep learning

KW - SUMO

KW - VANET

UR - http://www.scopus.com/inward/record.url?scp=85068963865&partnerID=8YFLogxK

U2 - 10.1109/VTCSpring.2019.8746657

DO - 10.1109/VTCSpring.2019.8746657

M3 - Conference contribution

BT - 2019 IEEE 89th Vehicular Technology Conference, VTC Spring 2019 - Proceedings

PB - Institute of Electrical and Electronics Engineers

CY - Piscataway

ER -

Wang Y, Menkovski V, Ho IWH, Pechenizkiy M. VANET meets deep learning: the effect of packet loss on the object detection performance. In 2019 IEEE 89th Vehicular Technology Conference, VTC Spring 2019 - Proceedings. Piscataway: Institute of Electrical and Electronics Engineers. 2019. 8746657. Available from, DOI: 10.1109/VTCSpring.2019.8746657