Model-Predictive Planning and Airspeed Regulation to Minimize Flight Energy Consumption
Liang Sun · 2025
Read Model-Predictive Planning and Airspeed Regulation to Minimize Flight Energy Consumption by Liang Sun (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-01311.txt.
Introduction
Trevor Karpinski and co-authors present a 2025 paper on saving flight energy in electric aircraft. Battery limits still block wide use of electric planes for cargo and passengers. Energy must be spent with care on each flight. Models of energy use keep getting sharper. Real-time control that follows those models is scarce. Take-off and landing energy are often left out of the math.
The team states core energy equations for the full flight. A model-predictive planner sets the plan along a fixed path. A classic PID controller then regulates airspeed in flight. Tests run in simulation. Operators get a control scheme that chases low energy use minute by minute. Even small savings per flight compound across a fleet, and they stretch range where batteries cannot grow.
Catalog record
| CREATOR | Liang Sun |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 40 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-01311
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Model-Predictive Planning and Airspeed Regulation to Minimize Flight Energy Consumption Liang Sun Abstract Although battery technology has advanced tremendously over the past decade, it continues to be a bottleneck for the mass adoption of electric aircraft in long-haul cargo and passenger delivery. The onboard energy is expected to be utilized in an efficient manner. Energy concumption modeling research offers increasingly accurate mathematical models, but there is scant research pertaining to real-time energy optimization at an operational level. Additionally, few publications include landing and take-off energy demands in their governing models. This work presents fundamental energy equations and proposes a proportional-integral-derivative (PID) controller. The proposed method demonstrates a unique approach to an energy consumption model that tracks real-time energy optimization along a predetermined path. The proposed PID controller was tested in simulation, and the results show its effectiveness and accuracy in driving the actual airspeed to converge to the optimal velocity without knowing the system dynamics. We also propose a model-predictive method to minimize the energy usage in landing and take-off by optimizing the flight trajectory.. This manuscript originated from the conference presentation in AIAA SciTech on January 23, 2023, at National Harbor, Maryland, USA. AIAA paper number: AIAA 2023-0303. 1 Introduction An unmanned aerial system (UAS) consists of an unmanned aerial vehicle and its supporting devices, typically autonomous, set out to complete a specified task. Multicopter UASs, also referred to as drones, have great utility due to their ability to fly over geographical barriers and have been utilized in various applications such as package delivery in urban environments [ 1 ] , search and rescue missions [ 2 ] , or combating building fires [ 3 ] . Nonetheless, there are unique considerations that need to be recognized to determine the comparative effectiveness over ground-based vehicles. For example, battery storage technology severely constrains the performance of both automotive and multicopter modes. Furthermore, realistic optimization of energy expenditure is made difficult by the complicated nature of multicopter dynamics. While modern literature covers energy expenditure estimation for drones, or path planning optimization in the Traveling Salesman problem, there appears to be little to no research with both fields in mind. Thus, long-term offline path planning that focuses on energy optimization combined with a real-time motion controller could provide a unique approach to modern problems of UAS applications. With these considerations, the energy-aware path planning and control of a drone seeks to greatly increase the utility and application of drones that would allow for greater mass adoption of UASs. Existing research for analysis of multicopter energy expenditure can be divided into two areas: estimation of energy usage, and energy optimization through path planning. The first area of research - calculating the energy expenditure of a drone over a time interval or at a specific point in time - is non-trivial and requires an in-depth analysis of the aerodynamic properties of a drone. Zhang et al. [ 4 ] conducted a meta-analysis on various energy consumption models, such as the ones developed by Kirchstein [ 5 ] , Dorling et al. [ 6 ] , D’Andrea [ 7 ] , and Liu et al. [ 8 ] , and categorized two approaches for energy expenditure calculation: those that assume lift-to-drag ratio to be the dominant factor, and those assuming that drones in forward flight will have similar energy expenditure formulation as a hovering helicopter. D’Andrea [ 7 ] falls under a model for energy expenditure that solely focuses on a critical parameter: the lift-to-drag ratio. It stems from a general-purpose analysis that seeks to approximate many classes of drones under the singular model, thus avoiding the lengthy analysis pertaining to each class of drone. An alternative integrated model was introduced in Figliozzi [ 9 ] with an aim to analyze energy expenditure for the purpose of approximating greenhouse gas emissions for environmental impact. While these models serve an important role, their general formulation may not be accurate enough for optimization purposes. Models introduced by Kirchstein [ 5 ] and Dorling et al. [ 6 ] fall under the category of component models, which are models with the base assumption that power requirements for a helicopter in forward flight, ascent, and descent are approximately equal to the power requirements while hovering. Liu et al. [ 8 ] assessed the accuracy of a component model by conducting field tests that demonstrate a 16.5% error from the theoretical calculations that under-estimates the real-world energy demands for forward flight. The component models provide a more comprehensive analysis of the dynamics of multicopters, and thus more accurately represent the energy demands of the UAS, but at the cost of being heavily reliant on dynamic properties of a specific drone, such as the coefficient of drag and blade-lift coefficient. After careful evaluation of the presented literature, the energy model formulated by Kirchstein [ 5 ] is the most comprehensive, and will be used as the primary model in the energy expenditure modeling and analyses in this paper, with contributions from Liu et al. [ 8 ] due to their empirical data evaluations. Samiei et al. [ 10 , 11 ] and Selje et al. [ 12 ] presented machine-learning-based approaches to estimating flight energy consumptions. However, these data-driven methods have demonstrated accurate energy consumption estimates only within the range covered by the data, while their predictive accuracy beyond this range remains unexplored. Calderon Ochoa et al. [ 13 ] presented an approach to estimating key pamaremters, such as the drag coefficient, of a multicopter for flight energy consumption modeling. The energy optimization approaches for UASs can be categorized into controller-based and algorithm-based methods. Controlled-based methods were evaluated in Okulski and Ławryńczuk [ 14 ] , who sought to improve drone maneuverability, by comparing a standard Model Predictive Control (MPC) method, a PID controller, and an MPC controller with the native algorithm augmented to a trajectory-guessing algorithm. They found that the augmented MPC controller performed about 15.7% better than the PID controller with regards to the attitude of the drone, which is the stability and control during flight, yet it consumed roughly 15.7% more energy than the PID controller (3.50Wh to 4.05Wh). It is noted that Okulski and Ławryńczuk [ 14 ] didn’t optimize energy directly, instead stating that attitude stabilization would lead to overall energy efficiencies. Our proposition lies within the context of reducing energy consumption specifically. Among algorithmic approaches to energy optimization, the work of Shivgan and Dong [ 15 ] utilizes a genetic algorithm to seek the minimum energy expenditure along a path. The genetic algorithm did prove to be more effective at optimizing the energy expenditure for waypoint path calculation versus a greedy algorithm, with a range of 9.6% to 81% reduced energy expenditure from 10 waypoints to 100 waypoints, respectively. Unfortunately, the Greedy algorithm has not been tested with real-time adjustments of the drone, but rather in path forecasting, and thus a different approach is required for this paper’s intended application of energy optimization achieved with real-time path correction. In addition, the research doesn’t have empirical backing to support the algorithm’s performance against external factors, which is incredibly disruptive in the realm of aerial flight, and we suggest our controller will be able to account for