Intelligent Aircraft Damage Assessment, Trajectory Planning, and Decision-Making under Uncertainty

Abstract

Situational awareness and learning are necessary to identify and select the optimal set of mutually non-exclusive hypothesis in order to maximize mission performance and adapt system behavior accordingly. This paper presents a hierarchical and decentralized approach for integrated damage assessment and trajectory planning in aircraft with uncertain navigational decision-making. Aircraft navigation can be safely accomplished by properly addressing the following: decision-making, obstacle perception, aircraft state estimation, and aircraft control. When in-flight failures or damage occur, rapid and precise decision-making under imprecise information is required in order to regain and maintain control of the aircraft. To achieve planned aircraft trajectory and complete safe landing, the uncertainties in system dynamics of the damaged aircraft need to be learned and incorporated at the level of motion planning. The damaged aircraft is simulated via a simplified kinematic model. The different sources and perspectives of uncertainties in the damage assessment process and post-failure trajectory planning are presented and classified. The decision-making process for an emergency motion planning and landing is developed via the Dempster-Shafer evidence theory. The objective of the trajectory planning is to arrive at a target position while maximizing the safety of the aircraft given uncertain conditions. Simulations are presented for an emergency motion planning and landing that takes into account aircraft dynamics, path complexity, distance to landing site, runway characteristics, and subjective human decision.