EcoFlight: Finding Low-Energy Paths Through Obstacles for Autonomous Sensing Drones
Jordan Leyva, Nahim J. Moran Vera, Yihan Xu, Adrien Durasno, Christopher U. Romero, Tendai Chimuka, Gabriel O. Huezo Ramirez, Ziqian Dong, Roberto Rojas-Cessa
TL;DR
EcoFlight addresses the problem of energy-constrained autonomous sensing drones navigating obstacle-rich 3D environments. It extends A* with an energy-aware cost that accounts for vertical ascent, horizontal movement, and acceleration, using a 3D map with 1 m spacing and an energy-scaled heuristic. The paper introduces the Rise and Traverse baseline for comparison and provides a detailed energy calculation framework that combines hover, drag, and acceleration into $E_{total}$. Experiments across obstacle densities from 30% to 75% show EcoFlight reduces energy consumption relative to obstacle-aware baselines, with larger gains at higher densities, and a suitable flight speed further amplifies savings. This work advances practical, energy-efficient autonomous sensing in cluttered airspace for urban and complex environments.
Abstract
Obstacle avoidance path planning for uncrewed aerial vehicles (UAVs), or drones, is rarely addressed in most flight path planning schemes, despite obstacles being a realistic condition. Obstacle avoidance can also be energy-intensive, making it a critical factor in efficient point-to-point drone flights. To address these gaps, we propose EcoFlight, an energy-efficient pathfinding algorithm that determines the lowest-energy route in 3D space with obstacles. The algorithm models energy consumption based on the drone propulsion system and flight dynamics. We conduct extensive evaluations, comparing EcoFlight with direct-flight and shortest-distance schemes. The simulation results across various obstacle densities show that EcoFlight consistently finds paths with lower energy consumption than comparable algorithms, particularly in high-density environments. We also demonstrate that a suitable flying speed can further enhance energy savings.