smartphone-based system for driver anger scale estimation using neural network on continuous wavelet transformation

Monitoring of the driver decreases accidents by reducing the risky behaviors and causes decreases the fuel consumption by preventing aggressive behavior. but this monitoring is costly due to built-in equipment. in this study, we propose a new model to recognize driving behavior by smartphone data without any extra equipment in the vehicles which is an important added value for smartphones. this recognition process is done in this paper based on the continuous wavelet transformation on accelerometer data. then these patterns are fed to multilayer perceptron neural network to extend the information extracted from the corresponding features. also the magnetometer sensor is used to detect the maneuvers through the driving period. results show the accuracy of the proposed system is near 80% for pattern recognition. driver scale based on a standard questionnaires regarding to driver angry scale (das), is also estimated by the proposed multilayer perceptron neural network with 3.7% errors in the average.