EmCoop: A Framework and Benchmark for Embodied Cooperation Among LLM Agents

2026-02-27

AI Summary

This work addresses the lack of fine-grained benchmarks for analyzing collaboration in embodied multi-agent systems, which hinders understanding how collaboration emerges, evolves, and influences task success. We propose the first embodied collaboration benchmark that supports arbitrary numbers of agents and flexible communication topologies. Our approach introduces a two-layer architecture that decouples cognitive reasoning from embodied interaction, leveraging large language models for natural-language communication and reasoning. Systematic experiments are conducted in two scalable environments. By introducing a generalizable, process-level metric for collaboration quality—going beyond conventional task success rates—we enable, for the first time, fine-grained and reproducible analysis of collaborative dynamics and failure modes across varying team sizes and task configurations.

Technology Category

• Multiagent Systems: Coordination and Collaboration
• Intelligent Robots: Embodied AI
• Cognitive Modeling & Cognitive Systems: Agent Architectures

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

Real-world scenarios increasingly require multiple embodied agents to collaborate in dynamic environments under embodied constraints, as many tasks exceed the capabilities of any single agent. Recent advances in large language models (LLMs) enable high-level cognitive coordination through reasoning, planning, and natural language communication. However, fine-grained analyses of how such collaboration emerges, unfolds, and contributes to task success in embodied multi-agent systems are difficult to conduct with existing benchmarks. In this paper, we introduce EmCoop, a benchmark framework for studying cooperation in LLM-based embodied multi-agent systems. Our framework separates a high-level cognitive layer from a low-level embodied interaction layer, allowing us to characterize agent cooperation through their interleaved dynamics over time. Given a cooperation-constrained embodied task, we propose generalizable, process-level metrics that diagnose collaboration quality and failure modes, beyond final task success. We instantiate our framework in two embodied environments that scale to arbitrary numbers of agents and support diverse communication topologies, and use these instantiations to demonstrate how EmCoop enables systematic analysis of cooperation dynamics across team sizes and task settings. The project web page can be found at: https://happyeureka.github.io/emcoop.

Problem

• embodied cooperation
• LLM agents
• multi-agent systems
• benchmark
• collaboration dynamics

Innovation

• embodied cooperation
• LLM agents
• multi-agent systems
• process-level metrics
• cognitive-embodied decoupling