Quadrotor Simulation Packages for Reinforcement Learning

Authors

V. Honcharenko
Kharkiv National University of Radio Electronics
https://orcid.org/0009-0002-0370-3361
V. Yesilevskyi
Kharkiv National University of Radio Electronics
https://orcid.org/0000-0002-5935-1505

Abstract

This publication reviews the most popular quadrotor simulations applied in reinforcement learning for UAV control, obstacle avoidance, path planning, and swarm behavior. We emphasize the importance of high-quality simulations that integrate visual and physical models to mimic real-world dynamics and sensor data accurately. Several popular quadrotor simulation platforms - AirSim, gym-pybullet-drones, Flightmare, Aerial Gym, RotorPy, and RotorS, are evaluated for their different capabilities and supporting reinforcement learning applications. Each package is analyzed for its strengths in rendering, physical accuracy, parallelization, and compatibility with reinforcement learning frameworks. These simulators provide valuable tools for developing and testing reinforcement learning algorithms.


Інформаційні системи та технології ІСТ-2024

Pages

43-46

Published

July 17, 2025

Details about this monograph

ISBN-13 (15)

978-966-659-354-5