Abstract:

Repairing damages in high-value parts using additive processes can be more efficient than using state-of-the-art high-skilled manual processes. We describe DamageEst, an efficient computational geometry framework for detecting and estimating the damage volume (DV) and the inner damage surface (IDS) using point cloud data (PCD) of damaged parts and their original 3D models. DamageEst identifies points in PCD on the IDS to reconstruct the IDS. It then encloses the reconstructed IDS and original part in a slightly scaled up background mesh, from which the DV is reconstructed using Boolean operations. DamageEst also enables targeted overestimation of damage for repair using additive manufacturing followed by milling to guarantee high surface quality. Prior methods scale exponentially in both time and memory, while DamageEst scales in polynomial time and memory. DamageEst enables precise identification and representation of damages with minimal human intervention.