In 2004, 5,190 people were killed due to a traffic accident involving a commercial motor vehicle (CMV), up from 4,793 people killed in 2001 (Traffic Safety Facts, 2004; Traffic Safety Facts, 2001). Driver drowsiness is an important issue to consider when discussing CMVs. According to the FMCSA, over 750 people are killed and 20,000 people are injured each year due to drowsy CMV drivers (as cited in Advocates for Highway and Auto Safety, 2001). Driver drowsiness is an important issue for CMV drivers for several reasons, including long work shifts, irregular schedules and driving long hours on interstates and highways with no scenic interruptions to help keep the driver alert. Because of these and other factors, including the high mileage exposure that CMV drivers face, drowsiness is an important issue in a CMV driver's occupation. There were two main goals to this research: 1) gain a better understanding of the time-related occurrences of drowsy-related critical incidents (i.e., crashes, near-crashes and crash-relevant conflicts), and 2) obtain drivers' opinions of the 2004 Revised Hours-of-Service regulations. To do this, recent data were used from a Field Operational Test conducted by the Virginia Tech Transportation Institute in which 103 participants drove in an instrumented heavy vehicle for up to 16 weeks; video data, and sensor data were collected from each participant. In addition, actigraph data was collected from 96 of the 103 participants. Each vehicle was instrumented with four video cameras to capture images of the drivers face, the forward roadway, and the adjacent lanes on each side of the truck. In addition, multiple sensors were installed in the vehicle in order to collect data such as the driver's speed, braking patterns and steering wheel movement. These data were combined to provide a complete picture of each driver's environment and behavior while they drove their normal routes. Data analysts reviewed the data for critical incidents (crashes, near-crashes, and crash-relevant conflicts) and determined a drowsiness level for each incident; these downiness levels were compared to drowsiness levels of baseline incidents (i.e., normal driving periods). The results show that drivers were more likely to have a drowsy-related critical incident between 2:00 pm and 2:59 pm. In addition to the video and sensor data, each driver was asked to fill out a subjective questionnaire regarding the revised HOS regulations. Drivers preferred the revised HOS regulations over the old HOS regulations and the number one item that was preferred in the revised HOS regulations is the 34-hour restart which allows drivers to restart their work week by taking off 34 consecutive hours. / Master of Science

Driver distraction has become a major concern for the U.S. Department of Transportation (US DOT).  Performance decrements are typically the result of driver distraction because attentional resources are limited, which are limited; fatigue and drowsiness limit attentional resources further.  The purpose of the current research is to gain an understanding of the relationship between mobile device use (MDU), fatigue, through driving time and time on duty, and drowsiness, through time of day and amount of sleep, for commercial motor vehicle drivers.  A re-analysis of naturalistic driving data was used to obtain information about the factors, MDU, safety-critical events (SCE), and normal driving epochs.  Odds ratios were used to calculate SCE risk for 6 mobile device use subtasks and each of the factors, which were divided into smaller bins of hours for more specific information.  A generalized linear mixed model and chi-square test were used to assess MDU for each factor and the associated bins.  Results indicated visually demanding subtasks were associated with an increase in SCE risk, but conversation on a hands-free cell phone decreased SCE risk.  There was an increase in SCE risk for visual manual subtasks for all bins in which analyses were possible.  Drivers had a higher proportion of MDU in the early morning (circadian low period) than all other times of day that were analyzed.  These results will be used to create recommended training and evaluate policy and technology and will help explain the relationship between MDU, fatigue, and drowsiness. / Master of Science

The development of technologies for preventing drowsiness at the wheel is a major challenge in the field of accident avoidance systems. Preventing drowsiness during driving requires a method for accurately detecting a decline in driver alertness and a method for alerting and refreshing the driver. As a detection method, the authors have developed a system that uses image processing technology to analyse images of the road lane with a video camera integrated with steering wheel angle data collection from a car simulation system. The main contribution of this study is a novel algorithm for drowsiness detection and tracking, which is based on the incorporation of information from a road vision system and vehicle performance parameters. Refinement of the algorithm is more precisely detected the level of drowsiness by the implementation of a support vector machine classification for robust and accurate drowsiness warning system. The Support Vector Machine (SVM) classification technique diminished drowsiness level by using non intrusive systems, using standard equipment sensors, aim to reduce these road accidents caused by drowsiness drivers. This detection system provides a non-contact technique for judging various levels of driver alertness and facilitates early detection of a decline in alertness during driving. The presented results are based on a selection of drowsiness database, which covers almost 60 hours of driving data collection measurements. All the parameters extracted from vehicle parameter data are collected in a driving simulator. With all the features from a real vehicle, a SVM drowsiness detection model is constructed. After several improvements, the classification results showed a very good indication of drowsiness by using those systems.

The development of technologies for preventing drowsiness at the wheel is a major challenge in the field of accident avoidance systems. Preventing drowsiness during driving requires a method for accurately detecting a decline in driver alertness and a method for alerting and refreshing the driver. As a detection method, the authors have developed a system that uses image processing technology to analyse images of the road lane with a video camera integrated with steering wheel angle data collection from a car simulation system. The main contribution of this study is a novel algorithm for drowsiness detection and tracking, which is based on the incorporation of information from a road vision system and vehicle performance parameters. Refinement of the algorithm is more precisely detected the level of drowsiness by the implementation of a support vector machine classification for robust and accurate drowsiness warning system. The Support Vector Machine (SVM) classification technique diminished drowsiness level by using non intrusive systems, using standard equipment sensors, aim to reduce these road accidents caused by drowsiness drivers. This detection system provides a non-contact technique for judging various levels of driver alertness and facilitates early detection of a decline in alertness during driving. The presented results are based on a selection of drowsiness database, which covers almost 60 hours of driving data collection measurements. All the parameters extracted from vehicle parameter data are collected in a driving simulator. With all the features from a real vehicle, a SVM drowsiness detection model is constructed. After several improvements, the classification results showed a very good indication of drowsiness by using those systems. / Title page is not included.

<p>Electrooculogram (EOG) data was used to develop, adjust and validate a method for drowsiness detection in drivers. The drowsiness detection was based on changes in blink behaviour and classification was made on a four graded scale. The purpose was to detect early signs of drowsiness in order to warn a driver. MATLAB was used for implementation. For adjustment and validatation, two different reference measures were used; driver reported ratings of drowsiness and an electroencephalogram (EEG) based scoring scale. A correspondence of 70 % was obtained between the program and the self ratings and 56 % between the program and the EEG based scoring scale. The results show a possibility to detect drowsiness by analyzing blink behaviour changes, but that inter-individual differences need to be considered. It is also difficult to find a comparable reference measure. The comparability of the blink based scale and the EEG based scale needs further investigation.</p>

Electrooculogram (EOG) data was used to develop, adjust and validate a method for drowsiness detection in drivers. The drowsiness detection was based on changes in blink behaviour and classification was made on a four graded scale. The purpose was to detect early signs of drowsiness in order to warn a driver. MATLAB was used for implementation. For adjustment and validatation, two different reference measures were used; driver reported ratings of drowsiness and an electroencephalogram (EEG) based scoring scale. A correspondence of 70 % was obtained between the program and the self ratings and 56 % between the program and the EEG based scoring scale. The results show a possibility to detect drowsiness by analyzing blink behaviour changes, but that inter-individual differences need to be considered. It is also difficult to find a comparable reference measure. The comparability of the blink based scale and the EEG based scale needs further investigation.

Approximately 1.35 million people die each year in car accidents and it is the most common cause of death for people aged 5-29. Because of this it is of large interest to be able to detect when a driver enters a sleepy state and to be able to alert the driver. This thesis aims to develop and evaluate a video-based system for driver sleepiness classification, and also to investigate personalised classification. The dataset used in this work consisted of face videos of 89 drivers captured using an IR camera. All drivers drove on a highway in real traffic in four different driving sessions, two at night time while being sleep deprived and two at day time while being alert. The drivers estimated their sleepiness level every 5th minute according to the 9-point Karolinska Sleepiness Scale (KSS).  The proposed driver sleepiness detection system consists of two steps. First, facial features are extracted from videos. Facial features corresponding to 60 seconds of data were used as input to a 1D-convolutional network. The network architecture consisted of five convolution blocks follower by a fully connected (dense) layer and an output layer with linear activation. The KSS-ratings were used as target values. For classification, the output was rounded to the closest KSS-level. KSS-levels 1-4 were merged into a single class, giving a total of 6 classes. The hyperparameters of the developed models were set by finding the best performing model during a grid search. Personalised models were developed by leaving out data from one of the drivers during training, and then finetuning the model parameters of the last dense layer by training on three of the driving sessions from the left out driver and evaluating on the fourth session. On average, the generic models of the developed architecture obtained an accuracy of 22%, a mean absolute error (MAE) of 1.49 and macro-F1 of 0.16. For the personalised models the mean results were an accuracy of 21%, MAE of 1.50 and macro-F1 of 0.14. The generic model showed signs of bias towards the majority classes (low KSS-levels) despite using weighting to compensate for class imbalance. Further, the results did not improve when personalising the model to compensate for individual differences. Future work should investigate if a larger and more balanced dataset is the only remedy, or if algorithm-level methods can be used to get around the problem.

This study was conducted in an attempt to improve drowsiness detection in automobiles by examining velocity-related measures. These measures were also included in multiple regression-generated drowsiness detection algorithms to determine their contribution to detection accuracy. In addition, the effects of cruise control and an auditory secondary task on the level of drowsiness and driving performance were examined. Twelve volunteers from the Blacksburg, Virginia area were used as subjects. In the early morning hours after sleep deprivation, subjects drove a moving base automobile simulator, during which time physiological and performance measures were gathered. Data analysis revealed that velocity-related measures can be good indicators of drowsiness when subjects are without external stimulation, but otherwise, these measures are fairly weak indicators of drowsiness. Similarly, the addition of velocity-related measures to drowsiness detection algorithms proved to be quite modest. Finally, there was no significant main effect of either cruise control or secondary task on drowsiness or driving performance. / M.S.

This on-road field investigation employed, for the first time, a completely automated, trigger-based data collection system capable of evaluating driver performance in an extended duration real-world commercial motor vehicle environment. The complexities associated with the development of the system, both technological and logistical and the necessary modifications to the plan of research are presented herein This study, performed in conjunction with an on-going three year contract with the Federal Highway Administration, examined the use of self-rating of alertness and temporal separation (minimum time-to-collision, minimum headway, and mean headway) as indicators of driver fatigue. Without exception, the regression analyses for both the self-rating of alertness and temporal separation yielded models low in predictive ability; neither metric was found to be a valid indicator of driver fatigue. Various reasons for the failure of self-rating of fatigue as a valid measure are discussed. Dispersion in the data, likely due to extraneous (non-fatigue related) factors (e.g., other drivers) are credited with reducing the sensitivity of the temporal separation indicators. Overall fatigue levels for all temporal separation incidents (those with a time-to-collision equal to or less than four seconds) were found to be significantly higher than for those randomly triggered incidents. On this basis, it is surmised that temporal separation may be a sensitive indicator for time-to-collision values greater than the 4-second criterion employed in this study. Two unexpected relationships in the data are also discussed. A "wall" effect was found to exist for minimum time-to-collision values at 1.9 seconds. That is, none of the participants who participated in this research effort exhibited following behaviors with less than a 1.9-second time-to-collision criterion. In addition, based upon the data collected for this research, anecdotal evidence suggests that commercial motor vehicle operators do not appear to follow the standard progression of events associated with the onset of fatigue. / Ph. D.

This study was undertaken to determine which variables and combination of variables could be used for the prediction of on-the-road drowsiness. Numerous driver-vehicle performance measures and secondary task performance measures were collected so that the predictability of several definitional measures of drowsiness could be tested. Twelve volunteer subjects were employed in the algorithm development phase of this study. All subjects were from the driver population in the Blacksburg, Virginia area. The participants were sleep deprived and drove a moving base simulator late at night in order to increase the likelihood that they would experience drowsiness while driving. After completion of data collection, numerous algorithms were developed using multiple regression and discriminant analysis methods. Another twelve volunteer subjects were subsequently employed in the algorithm validation phase of this study. Similar physiological and driving performance measures were collected during both phases of the study. All subjects were from the same driver population. All subjects were run under similar conditions as those in the algorithm development phase. Algorithms that appeared promising which were developed in the first phase of study were validated by applying them to the new data in an attempt to predict drowsiness on a new subject pool. It was found that drowsiness could be detected on a new subject pool and that the rate of correct predictions was quite high. There was no general decrease in predictive power of the drowsiness detection algorithms when applied to new data. Results showed that an accuracy rate of over 90 percent could be accomplished when output from the detection algorithms were classified into categories of "Awake," "Questionable," and "Drowsy." / Ph. D.