Identification of Driver Posture and Machine Learning Based Prediction

---

Thesis Type: Postgraduate

Institution Of The Thesis: Yildiz Technical University, Faculty of Mechanical Engineering, Department of Mechatronics Engineering, Turkey

Approval Date: 2020

Thesis Language: Turkish

Student: Ceyhun IBOLAR

Principal Supervisor (For Co-Supervisor Theses): Erhan Akdoğan

Co-Supervisor: Emre Çetin

Open Archive Collection: AVESIS Open Access Collection

Abstract:

Most of the traffic accidents are because of driver distraction that may be as a result of engagement in a secondary activity or the driver's state of mind (sleepy, tired, angry, etc.). There are many studies available in the literature about determining and monitoring the driver attention. Almost all academic studies and commercial products on some luxury intelligent cars on the market use vision based systems for this purpose. Furthermore, researches are ongoing to improve the algorithms that interpret the driver's physical measures. On the other hand, driver performance can be interpreted using the variables collected in intelligent cars by suitable algorithms. Some commercial products use the history of steering wheel torque and position variables to identify the driver distraction. With continually improving technology, sensors are evolving both in terms of precision and price. An example is the torque sensor which started to be used to measure the torque applied by the driver in order to determine the level of assistance of the steering system. Torque/force sensors can be used to interpret both the driver posture and the state of mind as the steering wheel is one of the most critical parts of the car where the driver interacts with. For instance, if the driver is maneuvering with single hand, the position of the driver hand can be determined analytically from the torques and forces that the driver exerts to the steering wheel. Hence, interpreting the driver's posture as well as state of mind may become possible using the hand position and the torques and forces that the driver exerts. In this study, a steering wheel jig is designed and produced equipped with a 6 axes torque/force sensor and an encoder in order to develop algorithms to determine