Markov Decision Process Contingency Landing Management Autonomy for Advanced Air Mobility

This paper presents a Markov Decision Process (MDP) based Contingency Management Autonomy (CMA) framework for electric vertical take-off and landing (eVTOL) aircraft with a decentralized battery-thruster propulsion system. A CMA Mission MDP is designed to identify low energy situations requiring a contingency landing. Based on reachability margin, the CMA MDP selects a Lift mode contingency vertiport landing given sufficient energy reserves and a more energy efficient Cruise mode runway landing otherwise. A modular simulation framework incorporating a multi-modal eVTOL surrogate model and waypoint and Dubins Cruise flight planner is developed to support real-time landing planning. The simulation computes battery end of discharge and reachability margin for use during CMA MDP policy execution. Results from a scenario involving onboard battery pack failures demonstrate the policy’s ability to select an energy-efficient runway landing that reduces total required energy by a factor of 18 compared to alternative vertical landing plans. This significant difference is due to the Cruise mode trajectory’s near-glide flight path angle leading to a stable 3 degree final runway approach. Results show the potential of CMA to enhance operational safety in future eVTOL operations.