Robots learn ownership to find personal items in everyday spaces
COOL: Curiosity-Driven Object Ownership Learning for Personalized Robotic Assistance
Robotics
Summary
Robots often struggle to know who owns which objects, which makes following commands like "Where is my backpack?" difficult. The authors developed COOL, a system that helps robots watch how people interact with objects over time to guess who owns what. COOL also remembers where things usually are and actively explores places to update this information. This helps robots find items based on ownership and deliver more personalized assistance.
What this means in practice
- •For home robotics developers: Enable robots to identify object ownership and locate personal items for improved personalized assistance at home.$Commercial implications: This paper enables consumer home robots to offer tailored help by recognizing who owns which objects, improving user satisfaction and utility.
- •For warehouse automation engineers: Improve robot systems to track item ownership or assignment in dynamic storage settings for better inventory handling.
Authors
Samira Huber, Ruben Hammele, Sören Pirk
Abstract
Robots are increasingly expected to provide personalized services in everyday environments. To do so, they must ground natural-language commands such as "Where is my backpack?" or "Find my bottle" and execute them by reasoning about object instances, people, locations, and ownership. This is challenging because ownership is rarely labeled explicitly and must be inferred from long-term, behavioral evidence of human-object interactions. To address this, we present COOL, a novel robotic framework for autonomously learning object ownership from everyday observations and maintaining a long-term spatial memory of its environment. To keep its memory current, COOL uses an agent-based curiosity-driven data collection strategy that guides the robot toward the most promising locations to gain information and refresh stale observations. Offline experiments, ablation studies, and real-world evaluations show that COOL can infer ownership relations from real-world interactions and use this knowledge for ownership-conditioned navigation and task execution.