An Efficient Moving Obstacle Avoidance Scheme for UAVs via Output Feedback Robust MPC

This article proposes an obstacle avoidance algorithm for unmanned aerial vehicles (UAVs) via output feedback robust model predictive control (MPC). The UAV is subject to unmeasurable states, additive state disturbances, and measurement noises, as well as state and control constraints. Efficient obstacle avoidance in the 3-D space is achieved for UAVs operating in challenging environments with both static and dynamic obstacles. To get estimations of unmeasurable states, we employ the Luenberger observer to estimate the state of the translational system of UAVs. For the 6-D UAV observer system and associated estimation error system, 6-D robust positively invariant sets are computed to tackle the uncertainties stemming from the bounded state disturbances, measurement noises, and state estimation errors, and thus, tightened state and control constraints are constructed for the nominal system. In addition, based on the time-varying position trajectory of moving obstacles, the separating hyperplanes for two time-varying convex sets are calculated and then a sequence of time-dependent safe sets is constructed by leveraging the concepts of safe sets and tubes. The inherently nonconvex moving obstacle avoidance constraints are then convexified into time-dependent closed polyhedral constraints. Thereby, the original nonconvex uncertain optimal control problem for UAVs is reformulated into a standard deterministic quadratic programming problem that can be solved efficiently. The proposed method also offers robust recursive feasibility and robust closed-loop stability of the controlled UAV system. The effectiveness of our proposal is validated through simulation results in a complex environment with multiple static and dynamic obstacles.