Interpretable Constructs for Human Object Arrangement Preferences
Analysis
This paper addresses the interpretability problem in robotic object rearrangement. It moves beyond black-box preference models by identifying and validating four interpretable constructs (spatial practicality, habitual convenience, semantic coherence, and commonsense appropriateness) that influence human object arrangement. The study's strength lies in its empirical validation through a questionnaire and its demonstration of how these constructs can be used to guide a robot planner, leading to arrangements that align with human preferences. This is a significant step towards more human-centered and understandable AI systems.
Key Takeaways
- •Identifies four interpretable constructs that explain human object arrangement preferences.
- •Validates these constructs through a self-report questionnaire.
- •Demonstrates the utility of these constructs by integrating them into a robot planner.
- •Results in robot arrangements that align with human preferences.
“The paper introduces an explicit formulation of object arrangement preferences along four interpretable constructs: spatial practicality, habitual convenience, semantic coherence, and commonsense appropriateness.”