CLiMRS: Cooperative Large-Language-Model-Driven Heterogeneous Multi-Robot System

Siqi Song, Xuanbing Xie, Zonglin Li, Yuqiang Li, Shijie Wang, Biqing Qi

Keywords: Heterogenous Robot Collaboration, Multi-LLM Planning, Robot Assembly

Abstract:

Cooperative multi-robot tasks often require heterogeneous agents to collaborate over long horizons while managing spatial constraints and execution uncertainties.
Although large language models (LLMs) excel at reasoning and planning, their potential for coordinated control in heterogeneous multi-robot teams has not been fully explored.
We present CLiMRS, a human-team-inspired adaptive negotiation paradigm that pairs each robot with an independent LLM agent and forms dynamic sub-groups for perception-driven discussions and cooperative planning under long-horizon uncertainty.
Within each group, local oracle planners lead parallel discussions to synchronize actions, while agents provide feedback to refine plans. This grouping–planning–feedback–execution loop enables efficient long-horizon planning and robust execution.
To evaluate these capabilities, we introduce CLiMBench, a heterogeneous multi-robot benchmark of challenging assembly tasks with diverse robot types and skill libraries.
Across both CLiMBench and a simpler benchmark, CLiMRS surpasses the best baseline, boosting success rates and improving efficiency by over 40% on complex tasks while maintaining very high success on simpler tasks.
Our results demonstrate that leveraging human-inspired group formation and negotiation principles markedly enhances the
efficiency of heterogeneous multi-robot collaboration.

Primary Area: applications to robotics, autonomy, planning

Submission Number: 523