NarraGuide: an LLM-based Narrative Mobile Robot for Remote Place Exploration
summary
NarraGuide is a working prototype: a mobile robot built on a TurtleBot base with mounted cameras, a SLAM map and a location-indexed tourbook feeding an LLM that performs LLM intent classification and generates location-aware narratives; a verifier sends navigation goals to the robot. Deployed in a geology museum with n = 20 participants, the system produces ~5 s average latency, supports dialogue and suggestions, yields roles like tour guide/companion/extension of self, and shows latency undermines fine-grained verbal control while bystander reactions range from playful to obstructive.
AI-written summary · about
chapters (14)
- 0:00Introduction to NarraGuide
- 0:45Telepresence robots overview
- 1:50LLM role in remote exploration
- 2:32Interaction design details
- 3:56Demo of robot suggestions
- 4:23Demo of user control
- 4:48Demo of exhibit queries
- 5:29System implementation
- 6:51User study setup
- 7:13Robot roles in user study
- 8:16Bystander reactions findings
- 9:39Bystander interaction demo
- 9:59Summary and job market note
- 10:20Q&A session starts
related talks
42mHow We Created a Thinking Robot?
09mNot All Who Wander Are Lost: A Localization-Free System for In-The-Wild Mobile Robot Deployments
34mLLM as a Robotic Brain: Cloud-Driven Robot Action Sequences Generated by Large...
32mROVINA
28mPreference Learning from Minimal Human Feedback for Interactive
55mUsing a Lidar for Robot Navigation in a Room
16mReal-World Robotics as the Next Frontier for AI?
18mLLMs and GenAI: A Guide to Building Your Own RAG Model for Document Interaction