Detection and Recognition of Archaeological Fragments to Improve Robotic Artifact Reconstruction

This thesis is part of the RePAIR project, which aims to develop a robotic platform to assist in the digitization and reconstruction of damaged archaeological artifacts. The research focuses on enhancing detection and recognition capabilities of the robotic platform, addressing two main tasks: detecting unknown fragments on a sandbed in the digitization phase and recognize fragments using a database of 3D models during the reconstruction phase. In order to achieve these objectives, baseline models are evaluated and improved using both synthetic and real data. The research connects the theoretical analysis with the practical application, including two testing phases on real-world experiments: an initial evaluation with 3D-printed fragments, followed by a field test with real-world data at the archaeological site of Pompeii. This thesis presents the work conducted up to the first phase, offering insights into the challenges and opportunities associated with the proposed approach.