A Dual-Approach Framework for eVTOL Climb Noise Mitigation

Technological advances in Urban Air Mobility (UAM) will bring new aircraft to the skies above metropolitan regions. With noise pollution emerging as a major barrier to public acceptance, mitigating the sound produced by electric vertical takeoff and landing (eVTOL) vehicles is essential to the future viability of UAM. This paper focuses on the climb phase of eVTOL operations, one of the loudest and most prolonged noise-generating segments of flight. We present two viable approaches for generating and evaluating climb trajectories with respect to both acoustic impact and energy use. The first uses direct collocation via the PSOPT (Problem Solving for Optimal Control) framework, an open-source software package for optimal control problems, to generate climb trajectories of a commercial quadrotor at varying gradients. These trajectories are then evaluated for Sound Exposure Level (SEL) and energy consumption using a simplified model in lieu of full-scale simulation tools; we report separate minima for noise exposure and energy consumption rather than a weighted multi-objective optimum. The second approach, developed in parallel by our team, outlines a deep reinforcement learning (DRL) framework to explore climb planning from a data-driven perspective. In this setup, a DRL agent, using Q-learning algorithms, is designed to adjust climb profiles based on feedback from SEL and energy metrics, offering a data-driven complement to the model-based approach. Together, these methods lay the groundwork for a modular, scalable framework to support noise- and energy-aware trajectory design in future eVTOL operations. While our paper outlines both methodologies, only preliminary results are presented for the optimal control approach, with DRL training and multi-objective weighting left for future work.