Reproduce “Human-Robot Dialogue Annotation for Multi-Modal Common Ground” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Claire Bonial, Stephanie M. Lukin, Mitchell Abrams, Anthony Baker, Lucia Donatelli, Ashley Foots, Cory J. Hayes, Cassidy Henry and 6 more
A multi-modal human-robot dialogue corpus, SCOUT, was developed and annotated with novel schemas for propositional meaning (Dialogue-AMR), dialogue structure, and visual context, providing empirical data for training autonomous systems to establish common ground in remote, constrained environments. The Dialogue-AMR annotations achieved an 86.6% Smatch score, demonstrating the scheme's ability to represent robot-actionable semantic content from natural language.
Key insights
- Dialogue-AMR successfully captures illocutionary force, temporal information, and normalized propositional content, enabling the distillation of robot-actionable behaviors from unconstrained natural language.
- The multi-floor dialogue structure provides a framework for understanding how intentions are translated, relayed, and negotiated across different conversational agents in complex human-robot teams.
- Human operators adapt their visual information requesting strategies over time, with systematic approaches like 'Degrees of rotation' correlating with higher task success in constrained environments.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.