The industry adoption of innovative safety-analysis approaches for autonomous vehicles : A case study exploring the industry adoption of model-based safety analysis

Abstract: In recent years, autonomous vehicles have been gaining traction as a technology that can revolutionize society and save approximately $800 billion per year in the U.S. alone. However, due to their complex and dynamic behavior, ensuring the safety of these vehicles has proven to be a challenge for traditional safety analysis methods. Model based safety analysis (MBSA), an application of model-based systems engineering (MBSE), has been proposed as an approach that can help alleviate some of these challenges. By capturing the behavior of the system in the form of a domain model, a number of new safety analysis methods such as simulations and formal verifications are made possible. Although this innovation was first proposed nearly two decades ago, it has not yet seen widespread use in the automotive industry, and there is currently a lack of research studying the barriers and drivers that affect its adoption. This thesis uses a case study approach and the diffusion of innovations theoretical framework to address this gap and understand the different factors that influence the adoption of MBSA in an organization. It also examines how current safety analysis practices are affected by autonomous vehicle technology from a broader perspective, given the emphasis on technical challenges in the existing literature. The findings of this thesis suggest that there is slow adaptation of safety analysis for autonomous vehicles, due to a lack of resource allocation. Additionally, the growing demand for demonstrating the safety of these systems is driven by the novel nature of the technology, the need for public acceptance, and the potential for increased liability in case of accidents. These- aspects, combined with numerous technological challenges, and the likely need for changes in working methods and new collaborations put further pressure on the organization. Regarding the adoption of MBSA, the findings indicate a high relative advantage but low awareness of the method among safety practitioners. Furthermore, the high perceived complexity of the method contributes to skepticism regarding its practical implementation. Further challenges include a high initial investment to implement the method, a lack of compatibility with the existing needs and ways of working within the organization, large-scale trial-related difficulties stemming from the need for a long project time frame, and problems observing the benefits of MBSA because of a lack of suitable comparison metrics. The findings of this thesis can be used to understand how to efficiently navigate the transition to autonomous vehiclesand to successfully implement new safety analysis approaches suitable for autonomous vehicles.