A Fault-Tolerant Architecture with Reconfigurable Path Planning Applied to an Unmanned Aerial Vehicle

Recent research has demonstrated the utility of reconfigurable flight control as a method to achieve fault-tolerance on unmanned rotorcraft. The reconfigurable flight controllers used in this paper employ neural networks to adapt to uncertain fault conditions as well as other disturbances. Despite success of these controllers, they cannot in most cases fully recover the maneuverability of the unimpaired vehicle. This paper develops a fault-tolerant control architecture that integrates reconfigurable flight control and reconfigurable path planning to augment the reliability and autonomy of an unmanned rotorcraft. The proposed three-tier hierarchical architecture is applicable to fixed and rotary wing aircraft. After the occurrence of a fault, reconfigurable path planning optimizes the vehicle's flight path to take full advantage of the performance of the active reconfigurable flight controller. The generation of optimal flight paths is posed as a linear program and solved in real-time. Simulation results demonstrate that reconfigurable path planning can significantly reduce the vehicle tracking error during maneuvers which exceed the capability of the reconfigurable controller.