A profile-driven dynamic risk assessment framework for connected and autonomous vehicles

Abstract

The Internet of Things has already demonstrated clear benefits when applied in many areas. In connected and autonomous vehicles (CAV), IoT data can help the autonomous systems make better decisions for safer and more secure transportation. For example, different IoT data sources can extend CAV's risk awareness, while the incoming data can update these risks in real-time for faster reactions that may mitigate possible damages. However, the current state of the art CAV research has not addressed this matter well enough. This paper proposes a profile-driven approach to manage IoT data in the context of CAV systems through a dynamic risk management framework. Unlike the current inflexible risk assessment strategies, the framework encourages more flexible investigation of risks through different risk profiles, each representing risk knowledge through a set of risk input considerations, assessment methods and optimal reaction strategies. As the risks change frequently with time and location, there will be no single profile that can cover all the risks that CAVs face on the road. The uses of different risk profiles, therefore can help interested parties to better understand the risks and adapt to various situations appropriately. Our framework includes the effective management of IoT data sources to enable the run-time risk assessment. We also describe a case study of using the proposed framework to manage the risks for the POD being developed in the Innovate UK-funded CAPRI project.