Visual servoing is a powerful control technique that leverages real-time image feedback from cameras to guide the motion of robotic systems. In classical visual servoing approaches, the control laws are designed based on hand-crafted image features and geometric models. While these methods have proven successful, their reliance on well-defined environments and known object models can introduce limitations in complex and dynamic scenarios. The integration of artificial intelligence (AI), like supervised and reinforcement learning techniques, offers the potential to significantly enhance the adaptability and robustness of visual servoing systems.
This thesis investigates the use of artificial intelligence (AI) to develop advanced visual servoing techniques for UAVs that can leverage learning-based approaches to guarantee docking with onboard RGB-D and IMU in unconventional scenarios, e.g., low or lost visibilty on the target potentially with the need to restart the task because of failure. This thesis falls within the scope of the European AutoASSESS project aimed at developing UAVs that can autonomously perform contact aerial inspection in ballast water tanks of cargo ships.
Expected outcomes:
- A new or improved AI-powered visual servoing framework for aerial robotics.
- Simulation-based results showcasing the performance and potential benefits of the developed visual servoing system.
- If feasible within the project timeframe, deployment and testing of the developed framework in real-world experiments.