Learning to Generate Unambiguous Spatial Referring Expressions for Real-World Environments
This addresses the challenge of clear object reference in real-world environments for robotics, representing a strong specific gain rather than a foundational advance.
The paper tackles the problem of generating unambiguous spatial referring expressions for human-robot interaction by proposing a two-stage deep learning approach, resulting in expressions that are ~30% more accurate and ~32% more preferred by people compared to the state-of-the-art.
Referring to objects in a natural and unambiguous manner is crucial for effective human-robot interaction. Previous research on learning-based referring expressions has focused primarily on comprehension tasks, while generating referring expressions is still mostly limited to rule-based methods. In this work, we propose a two-stage approach that relies on deep learning for estimating spatial relations to describe an object naturally and unambiguously with a referring expression. We compare our method to the state of the art algorithm in ambiguous environments (e.g., environments that include very similar objects with similar relationships). We show that our method generates referring expressions that people find to be more accurate ($\sim$30% better) and would prefer to use ($\sim$32% more often).