Development of Optimally Personalized Vehicle Control System and Situational Evaluation Metrics in Crash Imminent Situations

Download or read book Development of Optimally Personalized Vehicle Control System and Situational Evaluation Metrics in Crash Imminent Situations written by SeHwan Kim (Mechanical engineer) and published by . This book was released on 2019 with total page pages. Available in PDF, EPUB and Kindle. Book excerpt: Automotive research and technological development to date have enabled improvement in driving comfort, operational efficiency, motion stability and, most importantly, safety. An expectation for perfect or near-perfect vehicle system automation is increasing. However, the actual application of high-level technologies becomes more challenging as more advanced and complex technologies develop because the roadways in real life are comprised of countless uncertainties. The coexistence of automated vehicles and human-driven vehicles on roadways will be inevitable and drivers exhibit diversified driving habits and decision-making idiosyncrasies, behaving differently even in the same situation. As such, an appropriate understanding of human driver behavior in various driving situations would be beneficial. This dissertation is motivated by a research assumption that people’s driving behavior, even in crash imminent situations, can be predicted by analyzing a wide spectrum of daily driving data which also can be utilized in designing personalized control systems especially in crash imminent situations. This dissertation presents several applications of advanced control theories to replicate and to predict drivers’ longitudinal (i.e. speed control) and lateral (i.e. steering wheel control) control behaviors in various driving situations by utilizing the respective drivers’ historic driving data. In addition, optimally personalized control systems based on personalized situational evaluations in crash imminent situations are presented. Three test vehicles and a virtual reality driving simulator were used to collect driving data. In addition, extensive naturalistic driving data which include several crash and near-crash events were used for identifying driver characteristics. The simulation results showed that the proposed models are able to replicate driving data and predict individual driver’s preferred control inputs, and successfully control a vehicle that adapts to individual drivers in crash imminent situations. The contributions of this dissertation include: 1) an analysis of human driver behavior in various driving situations for driving characteristic identification; 2) development of a prediction model which is able to provide driver’s preferred control inputs; 3) development of a personalized crash-imminence detection model based on drivers’ historic driving data; 4) applications of control theories to develop optimally personalized control models in crash imminent situations; and 5) development of various metrics to evaluate driving situations for personalized decision-making. It is expected that the findings of this dissertation will benefit further research on vehicle stability and safety as well as more advanced vehicle technologies, particularly in personalization technologies.