Open Access

Martin Schörner

• Due to advances in technology, the use of autonomous flying robots for inspecting large structures has become increasingly popular in recent years. However, most applications focus on performing outdoor inspections. In the manufacturing sector, automated inspections of large structures performed by autonomous flying robots are rare due to the additional challenges, such as the complex environment and the unavailability of GPS signals. Nevertheless, the global economy's demand for efficient manufacturing requires the automation of inspection processes in this sector. This work introduces a modular relative path planning, perception and navigation system for flying robots performing inspections that relies only on onboard sensors and is based on model tracking. The system supports safe autonomous flight while avoiding collisions in complex and highly dynamic environments. The system includes path and trajectory planning algorithms based on either the targeted inspection of specificDue to advances in technology, the use of autonomous flying robots for inspecting large structures has become increasingly popular in recent years. However, most applications focus on performing outdoor inspections. In the manufacturing sector, automated inspections of large structures performed by autonomous flying robots are rare due to the additional challenges, such as the complex environment and the unavailability of GPS signals. Nevertheless, the global economy's demand for efficient manufacturing requires the automation of inspection processes in this sector. This work introduces a modular relative path planning, perception and navigation system for flying robots performing inspections that relies only on onboard sensors and is based on model tracking. The system supports safe autonomous flight while avoiding collisions in complex and highly dynamic environments. The system includes path and trajectory planning algorithms based on either the targeted inspection of specific Points of Interest on the assembly or coverage of the entire surface of the model which can be used interchangeably depending on the specific inspection task. The architecture is designed to allow for easy integration of existing, ground-based inspection and error detection tools. It's functionality is demonstrated through a number of prototypes based on case studies focusing on inspecting aircraft fuselage parts and a wind turbine. The prototypes are used to perform simulated and real-world proofs of concept and evaluations. The results demonstrate the system's ability to plan efficient inspection paths, precisely navigate relative to the inspected structures, and detect and avoid obstacles in real time, thus validating its effectiveness in the complex environments encountered inside manufacturing facilities.…