Scholarlink

Multi-robot platforms are a rapidly evolving area in robotics research, offering improved efficiency, robustness, and scalability for applications such as industrial automation and search and rescue. These platforms leverage multiple autonomous or semi-autonomous agents to perform complex tasks that would be challenging for a single system to execute effectively. In this paper, we introduce FlowProtocol, a lightweight rolling protocol integrated into a declarative, low-latency software platform designed by the authors. FlowProtocol provides a reliable foundation for developing real-time multi-agent systems, incorporating communication primitives and modular architectural components. We demonstrate the application of FlowProtocol by implementing a function-calling agent that delivers Large Language Model services. This agent operates within a peer-to-peer micro-agent swarm, coordinating a fleet of resource-harvesting robots through LLM-guided human interaction. We evaluate key aspects of the proposed architecture through a set of functional tests and present experimental results that benchmark FlowProtocol against the Robot Operating System, highlighting its suitability for real-time, communication-intensive multi-agent applications.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Improving Human-Swarm Interaction Through Speech Control and Peer-to-Peer Micro-agent Communication

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Giovanni De Gasperis,
  • Daniele Di Ottavio,
  • Sante Dino Facchini

摘要

Multi-robot platforms are a rapidly evolving area in robotics research, offering improved efficiency, robustness, and scalability for applications such as industrial automation and search and rescue. These platforms leverage multiple autonomous or semi-autonomous agents to perform complex tasks that would be challenging for a single system to execute effectively. In this paper, we introduce FlowProtocol, a lightweight rolling protocol integrated into a declarative, low-latency software platform designed by the authors. FlowProtocol provides a reliable foundation for developing real-time multi-agent systems, incorporating communication primitives and modular architectural components. We demonstrate the application of FlowProtocol by implementing a function-calling agent that delivers Large Language Model services. This agent operates within a peer-to-peer micro-agent swarm, coordinating a fleet of resource-harvesting robots through LLM-guided human interaction. We evaluate key aspects of the proposed architecture through a set of functional tests and present experimental results that benchmark FlowProtocol against the Robot Operating System, highlighting its suitability for real-time, communication-intensive multi-agent applications.