An Evasive Maneuvering Algorithm for UAVs in See-and-Avoid Situations

In this paper, we present an collision avoidance algorithm for unmanned aerial vehicles (UAVs) based on model predictive control. When a UAV encounters other aircraft that is estimated to approach closer than the minimum safety margin, the vehicle must execute an emergency evasive maneuver to avoid the impending collision at all cost. During this procedure, the unmanned vehicle must compute in real time a safe and plausible trajectory based on the collected information on the predicted future path of other vehicles. During the evasive maneuver, the trajectory generation and control problem is very stringent since the conflict-free trajectory must be plausible with respect to the given vehicle dynamics with limited control input. Therefore, in this research, we propose a model predictive control-based trajectory planner to satisfy the requirements listed so far due to its capability to explicitly address the control problem of constrained nonlinear dynamic systems. We consider a few scenarios involving nearby flying objects with various velocity and incident angle conditions. The proposed algorithm is validated in a head-on collision scenario using unmanned aerial vehicles.