This degree project investigates the suitability of using Behavior Trees (BT) as an architecture for the behavioral layer in autonomous driving. BTs originate from video game development but have received attention in robotics research the past couple of years. This project also includes implementation of a simulated traffic environment using the Unity3D engine, where the use of BTs is evaluated and compared to an implementation using finite-state machines (FSM). After the initial implementation, the simulation along with the control architectures were extended with additional behaviors in four steps. The different versions were evaluated using software maintainability metrics (Cyclomatic complexity and Maintainability index) in order to extrapolate and reason about more complex implementations as would be required in a real autonomous vehicle. It is concluded that as the AI requirements scale and grow more complex, the BTs likely become substantially more maintainable than FSMs and hence may prove a viable alternative for autonomous driving.

Aggressive driving is a phenomenon that has taken on tremendous significance in society. While the issue has been studied from perspectives of several disciplines, relatively few comprehensive empirical investigations have been conducted. This may be due, at least in part, to a scarcity of comprehensive theoretical works in the field, from which methodical research hypotheses could be derived. This paper reviews major extant theories of general aggression to offer a rationale for choosing a particular framework to apply to the topic of aggressive driving. The social-cognitive model of aggressive driving is recommended, as it takes into account wide-ranging cognitive, situational, and dispositional factors. Implications for future research are also considered.

Introduction: With many thousands of deaths still annually attributable to driving under the influence (DUI), it remains imperative that we continually address the problem of producing and sustaining effective countermeasures, and that we subject these efforts to empirical scrutiny. This article presents relevant findings from state-wide datasets. Results: A formula generating a potentially useful metric for assessing aspects of the DUI prosecutorial chain is presented, focusing on the rate of proactive DUI arrests. While in need of cautious interpretation due to issues of inherent inaccuracies in large databases, small numbers of crashes and/or arrests in multiple jurisdictions, and the lack of replication in other states, the analyses show no relationship between the level of DUI arrest activity and DUI-related crashes. This finding brings into question the efficacy of the many millions of dollars devoted each year to targeted DUI enforcement, as it is currently being implemented. Conclusions: Results are discussed in terms of developing adequate disincentives to DUI so as to raise general deterrence via dramatic increases in proactive DUI enforcement and then engaging in pervasive and persistent social marketing of such efforts to maximize the perception that arrest and punishment for DUI is always imminent, that penalties will be swift, certain, and severe. It is echoed that accurate data need to be collected at all levels of the DUI arrest and prosecution process in every jurisdiction within a state, so as to facilitate the empirical assessment of countermeasure efficacy in reducing alcohol-related crashes. Impact on Industry: Given that this work needs to be replicated, the impact on the traffic safety industry is potentially huge. The present data indicate that law enforcement efforts to further abate DUI-related crashes are apparently ineffective, though likely necessary to maintain reductions achieved in the 80s and early 90s. Thus, to attain additional systematic reductions, a dramatic increase in enforcement will be necessary as will a diversification of abatement efforts, including an increase in aggressive social marketing tactics to positively impact our traffic safety culture by making DUI universally unacceptable (for a discussion of this latter issue and on the use of positive reinforcement to change driver behavior, see Dula & Geller, 2007).

No description available.

No description available.

Autonomous vehicles will help society if they can easily support a broad range of driving environments, conditions, and vehicles. Achieving this requires reducing the complexity of the algorithmic system, easing the collection of training data, and verifying operation using real-world experiments. Our work addresses these issues by utilizing a reflexive neural network that translates images into steering and throttle commands. This network is trained using simulation data from Grand Theft Auto V~\cite{gtav}, which we augment to reduce the number of simulation hours driven. We then validate our work using a RC car system through numerous tests. Our system successfully drive 98 of 100 laps of a track with multiple road types and difficult turns; it also successfully avoids collisions with another vehicle in 90\% of the trials.

No description available.

Government reports indicate that, on average, more than 3000 people die due to distracted driving each year, accounting for nearly 10% of all fatal car crashes. Other reports claim that two-thirds of fatal car accidents result from aggressive driving. Previous research has been inconclusive regarding how personality impacts distracted and aggressive driving behaviors. Therefore, the goal of this current study is to fill the gap in the literature concerning the role that personality plays in distracted and aggressive driving behaviors. We also explored the role that distracted and aggressive driving behaviors played in accident involvement. A sample of (N=327) participants were recruited using social media and the UCF SONA System. They were asked to self-report their driving behaviors and personality traits by completing a series of online questionnaires (ADBQ, BFI, DBQ, DDQ, DEMO, and IPIP NEO PI-R). Using this data, bivariate correlations were run using the Pearson Correlation Coefficients to determine the role that personality (OCEAN) plays in distracted and aggressive driving behaviors. We used the DDQ and the IPIP NEO PI-R to evaluate the relationship between personality and distracted driving, and we found that personality traits: Agreeableness, Conscientiousness, Extraversion, and Neuroticism were all significant predictors of distracted driving. Openness was the only one of the five personality traits to have no significant correlation. We used the ADBQ and the IPIP NEO PI-R to assess the relationship between personality and aggressive driving, and we found the same four personality traits: Agreeableness, Conscientiousness, Extraversion, and Neuroticism were all significant predictors of aggressive driving. Openness was, again, the only one of the five personality traits to have no significant correlation. Backward regression analyses were performed to determine what caused these relationships. The regression analysis displayed trait subscales: Morality, Cooperation, Self Discipline, Activity Level, Excitement Seeking, Anger, Emotionality, and Liberalism, each significantly contributed to driver distraction. Another backward regression analysis reveals trait subscales: Morality, Self-Efficacy, Dutifulness, Self Discipline, Anger, and Artistic Interests, each significantly contributed to driver aggression.

No description available.

Machine perception is a key element for the research on autonomous driving vehicles. In particular, we focus on the problem of lane detection with a single camera. Many lane detection systems have been developed and many algorithms have been published over the years. However, while they are already commercially available to deliver lane departure warnings, their reliability is still unsatisfactory for fully autonomous scenarios. In this work, we questioned the reasons for such limitations. After examining the state of the art and the relevant literature, we identified the key methodologies adopted. We present a self-standing discussion of bird’s eye view (BEV) warping and common image preprocessing techniques, followed by gradient-based and color-based feature extraction and selection. Line fitting algorithms are then described, including least squares methods, Hough transform and random sample consensus (RANSAC). Polynomial and spline models are considered. As a result, a general processing pipeline emerged. We further analyzed each key technique by implementing it and performing experiments using data we previously collected. At the end of our evaluation, we designed and developed an overall system, finally studying its behavior. This analysis allowed us on one hand to gain insight into the reasons holding back present systems, and on the other to propose future developments in those directions. / Thesis / Master of Science (MSc)