Loading...
| Friend's email | |
| Your name | |
| Your email | |
| enter code | |
This page was sent successfuly
- Type of Document: Ph.D. Dissertation
- Language: Farsi
- Document No: 56544 (19)
- University: Sharif University of Technology
- Department: Computer Engineering
- Advisor(s): Hemmatyar, Ali Mohammad Afshin; Jafari Siavoshani, Mehdi; Moshiri, Behzad
- Abstract:
- Examining and improving driving behavior leads to a reduction in road accidents. Driver behavior is generally divided into two categories: aggressive and safe. Driving culture can be enhanced by identifying aggressive behavior and informing drivers directly. This information is used to assign safe drivers to the missions of transport fleet management companies and organizations. Also, this information is used by insurance companies or traffic police to apply discounts or fines. In any case, the detection and notification of aggressive behaviors would reduce accidents and improve the lives and mental safety of passengers and drivers. The detection of driving events is an introduction to the detection of drivers’ aggressive behavior, which is used in implementing the Advanced Driver Assistance System. These systems detect driving events with the help of in-vehicle sensors or smartphone sensors. Using in-vehicle sensors is not cost-effective due to limitations such as different vehicles’ Diagnostic Trouble Codes and the high cost of installing and maintaining equipment. Nowadays, due to the expansion of the smartphone's presence in every vehicle, sensors inside these phones are used to detect driving behaviors. This research aims to provide a solution to detect driving events and drivers' behavior by analyzing the signal information of smartphone sensors. According to the signal characteristics of the sensors, the performance of models such as recurrent neural network (LSTM), convolutional neural network (CNN), deep network (DN), and support vector machine (SVM) in detecting driving events and drivers’ behavior were evaluated. Finally, with the help of the ensemble learning method, a hybrid model was presented as a suggested solution. In the evaluation phase, an Android application collected data from mobile phone sensors. During one year, this application collected and analyzed the driving data of fifty drivers. Afterward, criteria such as accuracy, precision, recall, false positive rate, and specificity were used in the evaluation process, and the proposed model had the best performance among the models, with 97% accuracy
- Keywords:
- Recurrent Neural Networks ; Convolutional Neural Network ; Multi-Layer Perceptron (MLP) ; Support Vector Machine (SVM) ; Smart Phones Sensors ; Driving Behavior ; Driving Event ; Drivers Behavior
Digital Object List
-
محتواي کتاب
- view
Bookmark