Searching in Space and Time: Unified Memory-Action Loops for Open-World Object Retrieval

Evidence notes

• Service robots must retrieve objects in dynamic, open-world settings where requests may reference attributes ("the red mug"), spatial context ("the mug on the table"), or past states ("the mug that was here yesterday"). Existing approaches capture only parts of this problem: scene graphs capture spatial relations but ignore temporal grounding, temporal reasoning methods model dynamics but do not support embodied interaction, and dynamic scene graphs handle both but remain closed-world with fixed vocabularies.
• The work is explicitly tied to TIAGo, making the product binding high confidence.
• The artifact adds external research coverage for TIAGo beyond company-authored releases.