Extracting Semantics: LLM-Guided Automatic Population of Robot Ontology from URDF
Researchers have developed a preliminary method to automatically generate semantic abstractions for robots by converting URDF models into populated ontologies. This approach leverages Large Language Models (LLMs) to infer meaningful semantics from URDF file identifiers, which often lack explicit meaning. The pipeline prompts LLMs with concepts from an existing ontology to ensure semantic alignment and uses techniques like majority voting and validation to enhance reliability. Initial evaluations on various robot descriptions suggest this method can effectively bridge the gap between low-level robot descriptions and the structured knowledge representations needed for human-robot interaction. AI