Abstract
Aerial robots with cameras on board can be used in surface inspection to observe areas that are difficult to reach by other means. In this type of problem, it is desirable for aerial robots to have a high degree of autonomy. A way to provide more autonomy would be to use computer vision techniques to automatically detect anomalies on the surface. However, the performance of automated visual recognition methods is limited in uncontrolled environments, so that in practice it is not possible to perform a fully automatic inspection. This paper presents a solution for visual inspection that increases the degree of autonomy of aerial robots following a semi-automatic approach. The solution is based on human-robot collaboration in which the operator delegates tasks to the drone for exploration and visual recognition and the drone requests assistance in the presence of uncertainty. We validate this proposal with the development of an experimental robotic system using the software framework Aerostack. The paper describes technical challenges that we had to solve to develop such a system and the impact on this solution on the degree of autonomy to detect anomalies on the surface.
Highlights
The maintenance of certain infrastructures requires periodical inspections of surfaces to find defects as symptoms of potential problems due to, for example, structural imperfections
In order to be more effective in this task, it is desirable for aerial robots to have a higher degree of autonomy
This paper presents a solution for visual inspection that increases the degree of autonomy of aerial robots following semi-automatic approach
Summary
The maintenance of certain infrastructures requires periodical inspections of surfaces (e.g., the surface of a dam, the facade of a building, an indoor wall, etc.) to find defects (e.g., holes, fissures, mould, spots, humidity, etc.) as symptoms of potential problems due to, for example, structural imperfections. It may be more realistic to follow an approach based on a human-robot collaboration in which the operator delegates certain routine tasks to the robot, but the robot asks for assistance in the presence of uncertainty. In this domain, different techniques can be considered to provide autonomy to robots (e.g., path planning, computer vision, obstacle avoidance, visual alignment, coordination of multi-robot inspection, etc.). The paper describes our approach for automated visual recognition of defects, using an existing method that we extended to be used with operator assistance. The paper provides the dataset of images from surfaces that we constructed and used to evaluate the quality of the visual recognition method
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
