abstract: Driving a vehicle is a complex task that typically requires several physical interactions and mental tasks. Inattentive driving takes a driver’s attention away from the primary task of driving, which can endanger the safety of driver, passenger(s), as well as pedestrians. According to several traffic safety administration organizations, distracted and inattentive driving are the primary causes of vehicle crashes or near crashes. In this research, a novel approach to detect and mitigate various levels of driving distractions is proposed. This novel approach consists of two main phases: i.) Proposing a system to detect various levels of driver distractions (low, medium, and high) using a machine learning techniques. ii.) Mitigating the effects of driver distractions through the integration of the distracted driving detection algorithm and the existing vehicle safety systems. In phase- 1, vehicle data were collected from an advanced driving simulator and a visual based sensor (webcam) for face monitoring. In addition, data were processed using a machine learning algorithm and a head pose analysis package in MATLAB. Then the model was trained and validated to detect different human operator distraction levels. In phase 2, the detected level of distraction, time to collision (TTC), lane position (LP), and steering entropy (SE) were used as an input to feed the vehicle safety controller that provides an appropriate action to maintain and/or mitigate vehicle safety status. The integrated detection algorithm and vehicle safety controller were then prototyped using MATLAB/SIMULINK for validation. A complete vehicle power train model including the driver’s interaction was replicated, and the outcome from the detection algorithm was fed into the vehicle safety controller. The results show that the vehicle safety system controller reacted and mitigated the vehicle safety status-in closed loop real-time fashion. The simulation results show that the proposed approach is efficient, accurate, and adaptable to dynamic changes resulting from the driver, as well as the vehicle system. This novel approach was applied in order to mitigate the impact of visual and cognitive distractions on the driver performance. / Dissertation/Thesis / Doctoral Dissertation Applied Psychology 2017

The dynamic behavior of a vehicle is predominantly controlled by the forces and moments generated at the contact patch between the tire and the road surface. As a result, tire characteristics can dramatically change vehicle response, especially during maneuvers that yields the tires to reach to the limits of its adhesion capacity. To assist the driver in such cases and to prevent other possible instability scenarios, various vehicle control systems e.g. anti-lock brakes (ABS), stability controllers (ESP, ESC) or rollover mitigation schemes are introduced, which are generally known as active vehicle safety systems. Based on the above facts, one can easily come to the conclusion that to improve upon the current control algorithms developed for the technology in use; a vehicle control system design requires accurate knowledge of the tire states. This study proposes the use of a smart tire system that can provide information on momentary variation of tire features through the sensor units attached directly on the tire and develops control algorithms based on this information to assure the match-up between tire and controller dynamics. A prototype smart tire system was developed for field testing and for detailed analysis of its potential. Based on the collected prototype data, novel observer and controller schemes were developed to obtain dynamic tire state information and to improve vehicle handling performance. The proposed algorithms were implemented and evaluated using numerical analysis in Matlab/SimulinkR environment. For a more realistic simulation environment, vehicle models were integrated from Mechanical Simulations CarSimR® software suite. / Ph. D.

Active vehicle safety systems for road vehicles play an important role in accident prevention. In recent years, rapid developments have been observed in this area with advancing technology and electronic control systems. Active yaw control is one of these subjects, which aims to control the vehicle in case of any impending spinning or plowing during rapid and/or sharp maneuver. In addition to the development of these systems, integration and cooperation of these independent control mechanisms constitutes the current trend in active vehicle safety systems design. In this thesis, design methodology and simulation results of an active yaw control system for two axle road vehicles have been presented. Main objective of the yaw control system is to estimate the desired yaw behavior of the vehicle according to the demand of the driver and track this desired behavior accurately. The design procedure follows a progressive method, which first aims to design the yaw control scheme without regarding any other stability parameters, followed by the development of the designed control scheme via taking other stability parameters such vehicle sideslip angle into consideration. A two degree of freedom vehicle model (commonly known as &ldquo / Bicycle Model&rdquo / ) is employed to model the desired vehicle behavior. The design of the controller is based on Fuzzy Logic Control, which has proved itself useful for complex nonlinear design problems. Afterwards, the proposed yaw controller has been modified in order to limit the vehicle sideslip angle as well. Integration of the designed active yaw control system with other safety systems such as Anti-Lock Braking System (ABS) and Traction Control System (TCS) is another subject of this study. A fuzzy logic based wheel slip controller has also been included in the study in order to integrate two different independent active systems to each other, which, in fact, is a general design approach for real life applications. This integration actually aims to initiate and develop the integration procedure of the active yaw control system with the (ABS). An eight degree of freedom detailed vehicle model with nonlinear tire model is utilized to represent the real vehicle in order to ensure the validity of the results. The simulation is held in MATLAB/Simulink environment, which has provided versatile design and simulation capabilities for this study. Wide-ranging simulations include various maneuvers with different road conditions have been performed in order to demonstrate the performance of the proposed controller.