Investigating risk assessment and role of safety concerns in autonomous vehicle

Abstract

Self-driving vehicles are of critical importance to a future sustainable transport system, which is expected to become widespread around the world. The future transportation system includes more automated driving technologies and intelligent systems. Autonomous vehicles are widely expected to be in common usage in cities and transform the mobility system in the coming years. Autonomous vehicle technology is expected to be an integral part of our future transportation system. However, a substantial amount of risk is associated with self-driving vehicles which must be considered by decision-makers effectively. Given that automated driving technology and how it will interact with the mobility system are substantially risky, the risks involved in self-driving vehicles need to be addressed appropriately. The identified knowledge gap of the pre-literature review is that an overview of the identification which completely considers all types of risks related to self-driving vehicles does not exist. In answer to the knowledge gap, the purpose of the first study is to rank the risks in self-driving vehicles. As the widespread usage of autonomous vehicles is closer to becoming a reality, substantial consideration should also be paid to the extent to which individuals choose vehicular mobility tools. Some evidence highlights the fact that the perceived safety of self-driving vehicle has a major impact on the social acceptance and application of autonomous technology. This may help identify the extent to which people will adopt and use self-driving vehicles and the rate at which their safety concerns might be realized on the road. The purpose of the second study is to examine vehicle ownership models to better understand the adoption of vehicles by considering some factors such as liability issues, cost, safety, and environmental characteristics. Investigating potential drawbacks and attitudes towards safety concerns of AVs have substantial amount of importance for transportation planners and policy-makers. Thus, this thesis will seek to provide a deeper understanding of behavioral aspects associated with individuals' autonomous vehicle choices. Specifically, it is investigated the travelers' risk perception toward AV future as well as the analysis and prioritization of the risks associated with self-driving vehicles. Chapter 2 summarizes the literature review related to road safety and risk landscape of self-driving vehicles, an overview on adoption and ownership of autonomous vehicle and literature about hybrid and integrated fuzzy multi-criteria decision-making methods and its applications to real-world problems. New hybrid MCDM approaches, integrating the extended Fuzzy AHP, extended Fuzzy TOPSIS, and extended Fuzzy VIKOR methods, are proposed for the selection problem in this research study. The proposed hybrid methods are extended with Pythagorean fuzzy sets to handle the complex decision problem involving self-driving vehicles. The extended approach comprises uncertainty in decision-making problems by implementing Fuzzy Set Theory (FST). Additionally, each Decision Maker (DM) in the decision-making team has their individual importance within the team. Chapter 3 formulates the methodological contributions of this research to the literature. Risk prioritization is a complicated multi-criteria decision-making (MCDM) problem that requires consideration of multiple feasible alternatives and conflicting intangible and tangible criteria. This study addresses the prioritization of risks involved with self-driving vehicles by proposing new hybrid MCDM methods based on the Analytic Hierarchy Process (AHP), the Technique for order preference by similarity to an ideal solution (TOPSIS) and Vlse Kriterijumska Optimizacija I Kompromisno Resenje (VIKOR) under Pythagorean fuzzy environment. According to the review of the literature on risk landscape of self-driving vehicles, there has been no study of reviewing and prioritizing all types of risks associated with autonomous vehicles. Therefore, an expert team is asked to specify the risks and criteria for evaluation. By personal interview with 4 experts from academia and industry as well as literature review, eight risks involved in autonomous vehicles, 7 main criteria together with 10 sub-criteria are identified, the proposed methodology is employed in Chapter 4. Calculations of proposed methods gave similar results. Distance from Pythagorean fuzzy positive ideal solution (PIS) and negative ideal solution (NIS) for Pythagorean fuzzy TOPSIS (PF-TOPSIS), and the utility measure Si, the regret measure Ri and VIKOR-specific index Qi for Pythagorean fuzzy VIKOR (PF-VIKOR) are utilized for ranking the alternatives. Based on the results of analysis, Cyber Attack Risk involves the maximum amount of risk, and the consecutive risk is Reputational risk. The result of the proposed model is validated by performing sensitivity analysis. Sensitivity analysis is carried out to specify the impact of criteria weights on the decision- making process. Sensitivity analysis is conducted by changing the weight of the maximum group utility, and through exchanging the weight values of criteria for observing which criteria are most significant and how the criteria weights affect the autonomous vehicles' risk prioritization. To further validate the effectiveness and robustness of the proposed hybrid method computations, two different comparative analyses are performed in this study. The performance of proposed methodology with Pythagorean fuzzy sets is compared with those with ordinary fuzzy sets and it is revealed that the proposed method provides informative and reliable outcomes to better represent the impreciseness of decision making problems. AHP method is also performed to conduct comparative analysis. Based on the AHP analysis, cyber attack risk was found to be the first risk, and the consecutive risk is reputational risk. This result is similar to the one acquired by the proposed method. The remaining ranking shows different precedence. Chapter 5 and Chapter 6 investigate the role of safety concerns in autonomous vehicle ownership choice. Survey design, data collection, and description of data sets are indicated in Chapter 5. 1197 respondents were recruited from Istanbul, Turkey to complete the stated choice experiment through a web-based survey. Revealed Preference (RP) and socio-demographics data set, and the impact of socio-demographic characteristics on vehicular mobility tools are analyzed in detail. Then, an experimental design is conducted to respond to the research questions as clearly as possible. Research methodology and results of autonomous vehicle ownership choice model estimation are given in Chapter 6. An overview of the disaggregate choice model, multinomial logit model (MNL) specification, and mixed logit choice model (ML) specification, specifically the Heteroscedastic mixed logit model are discussed in this chapter. The methodologies to test the model significance, such as Log Likelihood Ratio, Pseudo Rho Square tests are also provided. Multinomial logit and Heteroskedastic mixed logit models were estimated to unravel users' preferences concerning the selection of autonomous vehicles with distinction among private, shared, and ride-hailing vehicles. Results indicate that the adoption of those vehicles varied concerning the aforementioned characteristics, and these findings could help decision-makers to develop a vehicular mobility system for future transportation dynamics. The policy makers need to deal with legal issues surrounding automated vehicles, specifically liability issues. In case of any at fault accidents, vehicle and software manufacturers may be liable instead of users. Shifting the burden on manufacturers would enable users to purchase automated vehicles as there would be less liability upon them. Reduced cost is another factor affecting whether consumers will choose autonomous vehicles. Therefore, service providers should implement strategies for cost reduction. When commuters are also given an incentive to install ride-hailing applications, they will prioritize automated vehicle production which will incentive consumers to purchase ride-hailing and shared vehicles. The motivation of this study was to provide more insights into the identification and prioritization of self-driving vehicles' risks which is an essential knowledge gap in the literature. The evaluation criteria determined by this study is novel as well, hence the research gap about some safety issues on self-driving vehicles will be filled through the current study. The results are significant as they shed light on the risk perception of the self-driving vehicle industry taking into account the risk factors. The finding of this study will enable decision-makers to take into consideration not only the hacking and malfunction factors but also the environmental, acceptance, and liability factors by containing the vagueness of experts' judgments. The study demonstrates that more sustainable and safer travel is achieved by considering risks. The results of this study also present an essential contribution that could be useful to transportation planners for constructing the future transportation system by considering the preference for vehicular mobility tools. The research findings will enable AV developers to consider liability issues, cost, and safety factors before releasing the different types of self-driving vehicles.