An Intelligent Power and Propulsion Management System for Hybrid-Electric Air Mobility

Abstract

The transition toward sustainable aviation is no longer a distant ambition, it is a global imperative. As urbanization accelerates and conventional transport networks strain under rising demand, Advanced Air Mobility (AAM) has emerged as a transformative solution, promising efficient, flexible, and low-emission transport in densely populated areas. At the heart of this vision lie hybrid-electric Vertical Take-Off and Landing (VTOL) aircraft, uniquely capable of combining the vertical agility of helicopters with the aerodynamic efficiency of fixed-wing flight. Yet, the widespread adoption of these systems remains hindered by critical limitations: volatile energy demands, limited onboard energy storage, unpredictable mission profiles, and the absence of intelligent energy management systems capable of adapting in real time to changing environmental and operational conditions.
This thesis addresses these barriers by proposing and implementing a novel Intelligent Power and Propulsion Management System (iPMS) tailored for hybrid-electric VTOL platforms. Using the custom-designed Falcon LP-0 UAV as a testbed, a comprehensive hybrid-electric energy architecture is developed, integrating batteries, hydrogen fuel cells, solar panels, and an internal combustion engine in a modular framework. At the core of this system is a Physics-Informed Multi-Agent Reinforcement Learning (PI-MARL) controller capable of dynamic power allocation, propulsion hybridization, and multi-objective optimization across complex missions.
The PI-MARL system was trained and validated in a simulation environment built around a real-world medical UAV delivery mission in Abu Dhabi, encompassing variable mission segments with distinct operational goals such as energy efficiency, emissions reduction, range maximization, and time-critical delivery. Compared to baseline rule-based control, the trained agents demonstrated consistent and significant improvements, including a 28.6% reduction in energy consumption, 24.8% increase in mission range, 21.4% faster completion time, and a 14.9% increase in energy safety margin. These gains were made possible through adaptive decision-making, hybrid power scheduling, and phase-specific objective balancing.
By bridging the gap between AI-based control and practical UAV energy management, this work advances the performance boundaries of hybrid-electric aircraft and lays the groundwork for certification-aligned intelligent control systems that can support the future of autonomous aerial mobility.

Date of Award	2025
Original language	American English
Supervisor	Roberto Sabatini (Supervisor)