Autonomous Guidance of an Aerial Robot Using 6-DoF Visual Control and Behavior Trees

The use of unmanned aerial vehicles in inspection and maintenance tasks reduces risks to human operators, in-creases the efficiency and quality of operations, and reduces the associated economic costs. Accurate navigation is crucial for automating these missions. This paper introduces an autonomous guidance scheme for aerial robots based on behavior trees that combine Position-Based Visual Servoing (PBVS) and waypoint following. The behavior tree dynamically adapts the robot navigation strategy depending on the required level of accuracy, accommodating to different types of missions and the presence of obstacles. As a result, the proposed navigation scheme enables complex missions to be performed in tight environments with significant levels of safety and robustness. All functional modules have been integrated into a behavior tree architecture, which provides robust, reactive, modular, and flexible frameworks to handle high-level tasks in complex missions. Validation using extensive software-in-the-loop simulations has demonstrated the robustness and accurate performance of the method, confirming its potential for real-world applications. An open-source implementation of our method is released under the GNU GPL3 license.