ORCA-A*: A Hybrid Reciprocal Collision Avoidance and Route Planning Algorithm for UAS in Dense Urban Areas

The rapid development of drones (or Unmanned Aerial Systems) and their potential deployment in urban areas poses a number of safety issues. Some degree of automation is most probably necessary to ensure that the UAS missions are safely and efficiently performed in urban environments. In a context where a large number of non-cooperative, non-communicative UAS would fly in dense urban areas, decentralized and autonomous approaches naturally come to mind. In such approaches, each agent would navigate among the buildings while avoiding the other traffic. ORCA (Optimal Reciprocal Collision Avoidance) is a state-of-the art geometric method for robot collision avoidance that could be used as a Detect & Avoid logic on-board UAS. It was initially designed for the 2D-motion of holonomic robots and requires some adaptation in order to be applied to flying objects in an urban environment. In particular, ORCA is a short-term collision avoidance that is not designed for path planning in a complex urban environment. In this study, we introduce a hybrid method combining ORCA with an A * path-planning algorithm and show that ORCA-A * significantly reduces the separation losses when compared with the baseline ORCA in artificial scenarios of dense UAS traffic.