Autonomous unmanned aerial vehicle exploration of unknown environment
Loading...
Date
2026-06-26
Authors
Δημητρίου, Ελένη
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
This thesis presents a reactive navigation framework for autonomous Unmanned Aerial Vehicle (UAV) exploration of unknown subterranean environments. The work is motivated by the increasing demand for aerial robotic systems capable of operating in GPS-denied, hazardous and unstructured settings – such as collapsed buildings, post-explosion mines and underground tunnels – where human presence is either impossible or life-threatening.
The core navigation strategy is based on the Artificial Potential Field (APF) method, selected for its real-time reactive capabilities, low computational complexity and ability to generate smooth, collision-free trajectories directly from the sensor data. A key contribution of this work is the introduction of a deepest-point attractive force formulation, which replaces a predefined goal location with the largest visible free-space region identified from a 3D LiDAR point cloud. This adaptation enables continuous exploration without any prior knowledge of the environment.
A cascaded PID control architecture is implemented to execute the reactive trajectories generated by the APF layer. This controller decouples position, altitude and yaw control from the inner attitude loop, ensuring that rapid obstacle avoidance maneuvers do not compromise flight stability. The control system is built upon a previously validated framework and refactored in Python to ensure modularity and ease of integration within the Robot Operating System (ROS) environment.
Simulation-based validation is carried out in Gazebo using two distinct environmental scenarios: a highly unstructured subterranean cave world and a semi-structured underground metro station. In both scenarios, the UAV successfully completes autonomous exploration without collisions, demonstrating stable flight, effective obstacle avoidance and reliable adaptation to marrow corridors, intersections and elevation changes.
Description
Keywords
Unmanned aerial vehicles (UAVs), Autonomous navigation, Artificial potential field (APF), Subterranean exploration, 3D LiDAR, Reactive navigation, Cascaded PID control, ROS, Gazebo simulation