LLM-Powered Multi-Agent Proactive Communication System for Embodied Intelligence
Abstract: We presents a novel multi-robot collaboration framework leveraging large language models (LLMs) for improved communication, planning, and execution. By integrating a centralized message pool and LLM-assisted decision-making, our system addresses limitations of existing multi-agent frameworks. Experiments in the MuJoCo simulation environment demonstrate significant improvements in task completion rates, communication effectiveness, and decision-making accuracy. Our proactive communication system reduces redundancy and enhances fault tolerance, enabling efficient handling of unexpected situations. Future work will focus on improving information synchronization and multi-system collaboration, further enhancing efficiency and scalability in complex environments.
Paper Type: Long
Research Area: Multimodality and Language Grounding to Vision, Robotics and Beyond
Research Area Keywords: Multimodality, cross-modal application
Contribution Types: NLP engineering experiment, Data analysis
Languages Studied: English
Submission Number: 5778
Loading