HOSPIT-LLM

Project Overview

HOSPIT-LLM is designed to advance social robotics in hospitality by creating a rich, real-world dataset of human–robot interactions. Using the Reachy humanoid robot, the project captures synchronized multimodal information—including RGB-D video, speech, gestures, and object interaction—to support the development of more adaptive, responsive, and context-aware robotic assistants in service environments.

Syndesis Contribution

• Development of multimodal AI workflows for hospitality-oriented human–robot interaction
• Integration of edge AI and LLM-based processing for natural language understanding and annotation
• Support for synchronized data capture across video, audio, gestures, and object recognition
• Contribution to semi-automated dataset annotation using fine-tuned language models and vision models
• Participation in the creation of an open dataset to support robotics and AI research in hospitality and service scenarios