Publications

Journal Articles

Fine-Grained Task Planning for Service Robots Based on Object Ontology Knowledge via Large Language Models

Published in June 11, 2024

We present the Fine-Grained Task Planning (FGTP) framework, which uses object ontology and Large Language Models (LLMs) to create detailed action plans. FGTP processes text and images to understand object attributes and uses logic-based reasoning to find alternatives if tasks fail. Tests in VirtualHome and with the TIAGo robot confirm its effectiveness.

Recommended citation: Li X, Tian G, Cui Y. Fine-Grained Task Planning for Service Robots Based on Object Ontology Knowledge Via Large Language Models[J]. IEEE Robotics and Automation Letters, 2024.
Download Paper

Conference Papers

Transformer-Based Relationship Inference Model for Household Object Organization by Integrating Graph Topology and Ontology

Published in June 30, 2024

We introduce a dataset of household objects’ properties and relationships, along with a graph-based model combining GAT and BERT to predict these relationships. Using a Transformer framework, our approach improves robots’ ability to organize objects by accurately predicting their relationships.

Recommended citation: None
Download Paper

Li Xiaodong

Publications

Journal Articles

Fine-Grained Task Planning for Service Robots Based on Object Ontology Knowledge via Large Language Models

Conference Papers

Transformer-Based Relationship Inference Model for Household Object Organization by Integrating Graph Topology and Ontology