Reproduce “Autonomous Unmanned Aerial Vehicle Navigation using Reinforcement Learning: A Systematic Review” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Fadi AlMahamid, Katarina Grolinger
This systematic review comprehensively surveys the application of Reinforcement Learning algorithms to autonomous Unmanned Aerial Vehicle navigation, classifying over 50 algorithms by environmental characteristics and outlining a structured selection process. It synthesizes current research on UAV control, obstacle avoidance, path planning, and flocking, also identifying prominent simulation tools and future research directions.
Key insights
- Autonomous UAV navigation tasks extensively utilize Deep Reinforcement Learning (DRL) for functions like flight control, obstacle avoidance using various sensors, global/local path planning, and multi-UAV flocking.
- The research community frequently applies a specific set of established RL algorithms (Q-learning, DQN, Double DQN, DDPG, PPO) to UAV navigation, while many newer, high-performing algorithms from other domains remain underexplored.
- UAV navigation simulation environments commonly integrate an RL agent, a UAV flight simulator (e.g., ROS, Microsoft AirSim), and a 3D graphics engine (e.g., Gazebo, Unreal Engine) for cost-effective and safe evaluation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.