Title: Multi-sensor Based Detection, Identification and Notification of Abnormal Driving


    Driving itself is a multi-dimensional entity and a lot of research has been focused on it’s modeling and characterization that could help make the transport system intelligent and ensure road safety. Monitoring of driving behavior primarily requires the velocity of vehicle, the angular displacement of the vehicle and distance from the surrounding traffic. Previous work in this field incorporates IMU sensors for detection of maneuvers and their classification on the basis of their severity but it suffers from a lot of false positives. Similarly, the use of camera for eye-movement detection and further prediction of lane changes is also a proposed solution in related literature. Moreover, GPS module has been a prominent device employed for determining the velocity of the car. However, this project integrates the useful information extracted from IMU sensors with the eye-gaze prediction based on the data obtained from camera for accurate maneuver detection. Furthermore, instead of determining the velocity with the help of GPS, it was retrieved from the vehicle’s speedometer itself through proper communication with the on-board diagnostic system of the vehicle. To further ensure a safe driving experience, a LIDAR was mounted on the vehicle for notifying the driver when the relative distance between the vehicle and the surrounding traffic exceeded safe limits. The pre-processing of raw data from sensors preceded the implementation of optimally recursive data processing algorithms such as kalman filter and suitable classifier such as SVM that could accurately discern aggressive driving event from normal driving. Once the aberrant behavior was detected, it was notified to the desired recipient through a text message.

Advisor:  Dr. Muhammad Tahir   

Group Members:

Name: Aaham Ejaz
Phone: 03214028769

Name: Hamad Ahmed
Phone: 03314475244

Name: Maryam Khalid
Phone: 03361410349

Name: Numair Saleem
Phone: 03324706270