Safe, Occlusion-Aware Manipulation for Online Object Reconstruction in Confined Spaces

Recent work in robotic manipulation focuses on object retrieval cluttered spaces under occlusion. Nevertheless, the majority of efforts lack an analysis conditions for completeness approaches or methods apply only when objects can be removed from workspace. This formulates general, occlusion-aware task, and safe reconstruction a confined space with in-place rearrangement. It proposes framework that ensures safety guarantees. Furthermore, algorithm, which is instantiation this abstract monotone instances developed evaluated empirically by comparing against random greedy baseline randomly generated experiments simulation. Even scenes realistic objects, proposed algorithm significantly outperforms baselines maintains high success rate across experimental conditions.