Driver behaviour analysis using topological features
Mahmoud, Mostafa Mahmoud Mohammad Hossn, Mohamed, Shady, Nahavandi, Saeid, Nelson, Kyle and Hossny, Mohammed 2016, Driver behaviour analysis using topological features, in SMC 2016 : Proceedings of the IEEE International Conference on Systems, Man, and Cybernetics, IEEE, Piscataway, N.J., pp. 1-6, doi: 10.1109/SMC.2016.7844736.
Attached Files
Name
Description
MIMEType
Size
Downloads
Title
Driver behaviour analysis using topological features
SMC 2016 : Proceedings of the IEEE International Conference on Systems, Man, and Cybernetics
Publication date
2016
Conference series
Systems, Man, and Cybernetics IEEE International Conference
Start page
1
End page
6
Total pages
6
Publisher
IEEE
Place of publication
Piscataway, N.J.
Summary
Driving behaviour prediction is a challenging problemdue to the nonlinearity of human behaviour. Linear andnonlinear techniques have been used to solve this problem, andthey provide good results presented in the performance of thecurrent autonomous cars. However, they lack the ability to adaptto abruptness that happens because of the human factor. In thispaper, we introduce a method to extract persistent homologybarcode statistics. These statistics are useful as a representativeof the driving process including the human behaviour. Humanfactor identification requires finding features that preserve certainproperties against scalability, deformation, and abruptness.Topological Data Analysis (TDA) using persistent homologyprovides these features for driver behaviour prediction. Wecaptured a driver’s head motion as an experimental behaviouralcue, combined it with captured simulated vehicle data (locationand velocities). Barcodes are extracted using JavaPlex, thenwe extracted descriptive statistics to show the significance ofthese barcode as features for driver behaviour prediction. Thecorrelation between the extracted features shows a promisingstart for a behavioural tracking applications using TDA.
Every reasonable effort has been made to ensure that permission has been obtained for items included in DRO. If you believe that your rights have been infringed by this repository, please contact drosupport@deakin.edu.au.