Centralized LLM-Driven Multi-Robot Coordination for Cooperative Object Transportation

This paper proposes a centralized Large Language Model (LLM)-based Multi-Robot System (MRS) for application of cooperative object transportation. We integrate the LLM into a ROS-based MRS to improve structured task allocation and execution. By leveraging prompt engineering, natural language instructions are converted into hierarchical JSON commands, allowing the central host to efficiently parse and distribute tasks among robots for coordinated multi-robot control. In addition, to achieve efficient and adaptive navigation in different environments, the system integrates $A^{*}$ and Dynamic Window Approach (DWA) algorithms for global path planning and local obstacle avoidance. Simulation and real-world experiments validate the effectiveness of the proposed LLM-based MRS control method. This study contributes to the advancement of MRS in industrial and service applications.