Conflict detection and resolution algorithms for UAVs collision avoidance

Abstract Collision-avoidance is a safety-critical requirement to operate UAVs in non-segregated airspaces. In case of communication problems between a UAV and the corresponding pilot-in-command, a technology is required onboard the UAV to implement a capability to detect and avoid collision-hazards even autonomously. After an introduction to the problem of developing a so-called sense-and-avoid system and its avoid-function, this work presents a solution in terms of algorithms to implement the above capability. To detect and resolve potential mid-air conflicts, a geometric deterministic approach has been utilised: an intruder is modeled trough a moving-ellipsoid and a four-dimensional approach in the time-space domain provides the solution. The approach makes use of kinematics information to detect potential conflicts and to provide actions for conflict resolution, such as speed-changes in intensity and/or direction. The proposed solution also enables the UAV to meet the applicable vertical and horizontal minima of separation and to comply with real-time constraints.