Using machine learning and object detection for signal-controlled intersections

Harry Spaabæk Lahrmann; Andreas Berre Eriksen; Jakob Haahr Taankvist; Mikkel Færgemand Hansen; Kim Guldstrand Larsen

Using machine learning and object detection for signal-controlled intersections

Harry Spaabæk Lahrmann, Andreas Berre Eriksen, Jakob Haahr Taankvist, Mikkel Færgemand Hansen, Kim Guldstrand Larsen

Publikation: Bidrag til bog/antologi/rapport/konference proceeding › Konferenceartikel i proceeding › Forskning › peer review

Abstract

There is an increase in traffic volumes and, as such, a requirement for maximisation of the road capacity. It is crucial that there is awareness of the traffic volumes in order to make the right choices regarding road development. Video registrations and related software to video analysis to measure traffic volumes are increasingly used, but there is limited documentation on the reliability of these. This paper compares manual registrations, treated as ground truth, the hardware independent software RUBA and an on-the-shelf product. While the RUBA software, in general, had a reasonable precision on the direction parallel to the camera direction (8% and 3% deviations, respectively); it was less precise regarding transversal-driving vehicles (23% deviation). The on-the-shelf hardware had a significantly higher deviation regarding the two parallel directions, (35% and 67% deviations, respectively) and a reasonable deviation regarding transversal-driving vehicles (11% deviation). It indicates that on-the-shelf hardware might need further calibration in general.

Originalsprog	Engelsk
Titel	ITS 2018 Conference Proceedings : Transport network operations
Antal sider	10
Vol/bind	6
Udgivelsessted	Copenhagen
Forlag	ITS World
Publikationsdato	2018
Artikelnummer	EU-TP1618
Status	Udgivet - 2018
Begivenhed	25th ITS World Congress - Quality of Life - Bella Center, Copenhagen, Danmark Varighed: 17 sep. 2018 → 21 sep. 2018 https://itsworldcongress.com/

Konference

Konference	25th ITS World Congress - Quality of Life
Lokation	Bella Center
Land/Område	Danmark
By	Copenhagen
Periode	17/09/2018 → 21/09/2018
Internetadresse	https://itsworldcongress.com/

Emneord

Signal-controlled intersection
Machine learning
Object detection

AUB Link

Søg efter materialet i Aalborg Universitetsbiblioteks søgemaskine

Citationsformater

@inproceedings{edee491a0a2b42ad8799ec0152c466f5,

title = "Using machine learning and object detection for signal-controlled intersections",

abstract = "This paper presents new principles for managing signal-controlled intersections. By using machine learning and object detection as substitutes for point detection and offset between intersections, a controller for signal-controlled intersections has been developed using the optimization program UPPAAL. Using the micro simulation program VISSIM, the controller has been tested in four signal-controlled and coordinated intersections in the street Hobrovej in Aalborg, Denmark. The simulation shows that in comparison with the existing controller, this controller provides a reduction of between 30% and 50% in average delays, queues and number of stops. The fuel consumption and total travel time of cars in the coordinated section are reduced by about 20% in the simulation study.",

keywords = "Signal-controlled intersection, Machine learning, Object detection, Signal-controlled intersection, Machine learning, Object detection",

author = "Lahrmann, {Harry Spaab{\ae}k} and Eriksen, {Andreas Berre} and Taankvist, {Jakob Haahr} and Hansen, {Mikkel F{\ae}rgemand} and Larsen, {Kim Guldstrand}",

note = "Paper presented in SP7 - Data and information; 25th ITS World Congress - Quality of Life ; Conference date: 17-09-2018 Through 21-09-2018",

year = "2018",

language = "English",

volume = "6",

booktitle = "ITS 2018 Conference Proceedings",

publisher = "ITS World",

url = "https://itsworldcongress.com/",

}

Using machine learning and object detection for signal-controlled intersections. / Lahrmann, Harry Spaabæk; Eriksen, Andreas Berre; Taankvist, Jakob Haahr et al.
ITS 2018 Conference Proceedings: Transport network operations. Bind 6 Copenhagen: ITS World, 2018. EU-TP1618.

Publikation: Bidrag til bog/antologi/rapport/konference proceeding › Konferenceartikel i proceeding › Forskning › peer review

TY - GEN

T1 - Using machine learning and object detection for signal-controlled intersections

AU - Lahrmann, Harry Spaabæk

AU - Eriksen, Andreas Berre

AU - Taankvist, Jakob Haahr

AU - Hansen, Mikkel Færgemand

AU - Larsen, Kim Guldstrand

N1 - Paper presented in SP7 - Data and information

PY - 2018

Y1 - 2018

N2 - This paper presents new principles for managing signal-controlled intersections. By using machine learning and object detection as substitutes for point detection and offset between intersections, a controller for signal-controlled intersections has been developed using the optimization program UPPAAL. Using the micro simulation program VISSIM, the controller has been tested in four signal-controlled and coordinated intersections in the street Hobrovej in Aalborg, Denmark. The simulation shows that in comparison with the existing controller, this controller provides a reduction of between 30% and 50% in average delays, queues and number of stops. The fuel consumption and total travel time of cars in the coordinated section are reduced by about 20% in the simulation study.

AB - This paper presents new principles for managing signal-controlled intersections. By using machine learning and object detection as substitutes for point detection and offset between intersections, a controller for signal-controlled intersections has been developed using the optimization program UPPAAL. Using the micro simulation program VISSIM, the controller has been tested in four signal-controlled and coordinated intersections in the street Hobrovej in Aalborg, Denmark. The simulation shows that in comparison with the existing controller, this controller provides a reduction of between 30% and 50% in average delays, queues and number of stops. The fuel consumption and total travel time of cars in the coordinated section are reduced by about 20% in the simulation study.

KW - Signal-controlled intersection

KW - Machine learning

KW - Object detection

KW - Signal-controlled intersection

KW - Machine learning

KW - Object detection

M3 - Article in proceeding

VL - 6

BT - ITS 2018 Conference Proceedings

PB - ITS World

CY - Copenhagen

T2 - 25th ITS World Congress - Quality of Life

Y2 - 17 September 2018 through 21 September 2018

ER -

Using machine learning and object detection for signal-controlled intersections

Abstract

Konference

Emneord

AUB Link

Fingeraftryk

Citationsformater