Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • harvard1
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Automatic Classification of Traffic Situations using On-board Recorded Vehicle Data
KTH, School of Engineering Sciences (SCI), Aeronautical and Vehicle Engineering, Vehicle Dynamics.
2011 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

In this Master’s thesis a method to automatically classify log files containing recorded vehicle data from Scania trucks is investigated.The files were recordings of data from the on-board controller area network(CAN)-buses and were gathered from field test vehicles during normal operating conditions. All the recordings were triggered by the embedded collision warning system. The test vehicles were equipped with a forward looking radar giving information about the surrounding traffic. Seven classes of traffic situations were proposed as classes to try to detect automatically. These classes were Queue, Vehicle ahead turns left,Vehicle ahead turns right, Roundabout, Catch up, Overtake and Overtakebicycle.Using Matlab, an evaluation tool was made to handle and analyse the log files. The tool converts the log files in all necessary steps in orderto analyse and classify the contents of each file. The results is presented in plots and also stored in a file for later review.To classify the files automatically, three methods have been tested and two of the methods have been used for this task. The first method uses a set of conditions on the signals to determine the class, the secondmethod uses a decision tree to differentiate the classes. A cluster analysisof the log files showed not to be an effective method of classifying thefiles and was therefore not fully developed.The result showed that the classification rules detected over 85%of the 192 available situations compared to the “decision tree”-methodwhich detected a maximum of 75% of the situations. The drawback ofthe first method was that the more correct detections also meant morefalse detections. The specificity was used as a measure of how well the method was able to leave the non-existing situations undetected. Thespecificity for the “classification rules”-method was approximately 85%and for the decision tree the specificity was approximately 95%.The conclusion is that both the “classification rules”- and the “decisiontree”-method can be used to classify the log files, however thefirst method produces more detections but less specific, and vice versafor the latter. The methods can be further developed using more of theavailable signals or by working with the specific conditions used by themethods.

Place, publisher, year, edition, pages
2011. , 47 p.
Series
Trita-AVE, ISSN 1651-7660 ; 2011:32
National Category
Vehicle Engineering
Identifiers
URN: urn:nbn:se:kth:diva-89898OAI: oai:DiVA.org:kth-89898DiVA: diva2:503959
Subject / course
Vehicle Engineering
Educational program
Master of Science - Vehicle Engineering
Uppsok
Technology
Examiners
Available from: 2012-02-17 Created: 2012-02-17 Last updated: 2012-02-17Bibliographically approved

Open Access in DiVA

No full text

By organisation
Vehicle Dynamics
Vehicle Engineering

Search outside of DiVA

GoogleGoogle Scholar

urn-nbn

Altmetric score

urn-nbn
Total: 130 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • harvard1
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf