Risk assessment for sUAS in urban environments: a comprehensive analysis, modelling and validation for safe operations

The rapidly growing number of applications for small Unmanned Aerial Systems (sUAS) in last-mile applications within metropolitan environments creates a complex airspace and ground safety scenario, where numerous risks must be considered to accomplish safe operations. The built-up and heterogeneously shaped geometry, together with the densely populated and transited nature of urban scenes, define a challenging scene for piloted and autonomous missions, where operators need to consider performance-based and third-party risks. In response to the increasing requirements inherent from urban scenarios, this paper proposes an integrated and comprehensive risk model for urban sUAS operations, composed by different risk layers designed for real-world scenarios, and validated through simulated drone flights and 3D risk-based navigation. By identifying the different risk requirements for sUAS operations, first party risks – including navigation performance, data link monitoring and collision avoidance – are computed within a photorealistic simulation environment for a discretized airspace representation. On top of this, trajectory based third party risks are modelled to identify potential routes subject to drone failure and consequent fatality and third party damage, as well as societal impact in terms of noise and privacy. Risk-based navigation techniques are implemented to validate the resulting model, including classical path planning and reinforcement learning. The results enable the perception of urban scenes associated risks through the lenses of risk modelling, providing a valuable methodology for sUAS urban operations and contributing to safer drone flights.