Title: Third Party Risk Modelling and Assessment for Safe UAV Path Planning in Metropolitan Environments

Abstract: Various applications of advanced air mobility (AAM) in urban environments facilitate our daily life and public services. As one of the key issues of realizing these applications autonomously, path planning problem has been studied with main objectives on minimizing travel distance, flight time and energy cost. However, AAM operations in metropolitan areas bring safety and society issues. Because most of AAM aircraft are unmanned aerial vehicles (UAVs) and they may fail to operate resulting in fatality risk, property damage risk and societal impacts (noise and privacy) to the public. To quantitatively assess these risks and mitigate them in planning phase, this paper proposes an integrated risk assessment model and develops a hybrid algorithm to solve the risk-based 3D path planning problem. The integrated risk assessment method considers probability and severity models of UAV impact ground people and vehicle. By introducing gravity model, the population density and traffic density are estimated in a finer scale, which enables more accurate risk assessment. The 3D risk-based path planning problem is first formulated as a special minimum cost flow problem. Then, a hybrid estimation of distribution algorithm (EDA) and risk-based A* (named as EDA-RA*) algorithm is proposed to solve the problem. To improve computational efficiency, k-means clustering method is incorporated into EDA-RA* to provide both global and local search heuristic information, which formed the EDA and fast risk-based A* algorithm we call EDA-FRA*. Case study results show that the risk assessment model can capture high risk areas and the generated risk map enables safe UAV path planning in urban complex environments.