Özet:

Advanced Driving Assistants Systems (ADAS) have an important milestone for unmanned vehicles. The main goal of this study is to compare the performances of major classification algorithms for aggressive driving detection, which is one of the fundamental problems of ADAS, through CAN (Control Area Network) bus sensor data. Supervised Learning based Classification Algorithms (SLCAs) are employed by this study, which are Artificial Neural Networks (ANN), Support Vector Machines (SVM), K-Nearest Neighbors (KNN), C4.5 Algorithm (J48), and Naïve Bayesian Classifier. These algorithms utilize CAN bus (Controller Area Network Bus) data acquired by OBDII (On-board Diagnostics) socket of the vehicle to detect driver mood associated with driving style. With the aim of ground truth, aggressive and calm drive were tried by different subject drivers, and acquired CAN bus sensor data in question is labeled as "aggressive" and "calm”, in our experiments. Afterwards, these are normalized for proper modality in mentioned classification algorithms. In the end of the process, combining latter and former steps are transformed into training data to assess performances of SLCAs for driver mood detection. Resultant performance evaluation for the algorithms suggest that the Naïve Bayes Classifier is more successful than the others.

Anahtar Kelimeler: