LLawCo: Learning Laws of Cooperation for Modeling Embodied Multi-Agent Behavior
Qinhong Zhou, Chuang Gan, Anoop Cherian
Why It Matters
What makes this one worth your time
Improving cooperation in multi-agent systems can lead to more efficient and successful task completion in decentralized and partially observable environments.
LLawCo enhances multi-agent cooperation by learning and applying behavioral laws.
Summary
The paper introduces LLawCo, a framework designed to improve cooperation among embodied multi-agent systems by learning and applying behavioral laws. It evaluates the framework using a new benchmark, PARTNR-Dialog, and reports improvements in task success rates over existing methods.
Key contributions
- Introduction of the LLawCo framework for learning cooperation laws.
- Development of the PARTNR-Dialog benchmark for evaluating multi-agent cooperation.
- Demonstrated improvement in task success rates over existing frameworks.
Notable insights
- The use of past failures to derive behavioral laws for agents is a novel approach.
- Explicitly incorporating behavioral laws into agents' reasoning processes via fine-tuning is a clever methodology.
Possible limitations
- Not stated in the abstract
Abstract
arXiv:2606.28182v1 Announce Type: cross Abstract: Embodied agents operating in decentralized and partially observable environments have attracted growing attention in recent years. However, existing large language model (LLM)-based agents often exhibit behaviors that are misaligned with their partners or inconsistent with the environment state, leading to inefficient cooperation and poor task success. To address this challenge, we propose a novel framework, Learning Laws of Cooperation (LLawCo), that enables embodied agents to autonomously align with both their partners and task objectives. Our framework allows agents to reflect on past failures to extract misaligned behavioral patterns, which are used to derive high-level behavioral laws, such as "Talk when necessary" and "Wait for partner." These laws are explicitly incorporated into the agents' chains of thought via supervised fine-tuning, aligning their reasoning with task requirements and the behavior of other agents. To evaluate our approach, we introduce PARTNR-Dialog, a large-scale multi-agent communicative and cooperative planning benchmark built on the PARTNR environment. Experiments on existing tasks and our new benchmark demonstrate significant improvements in cooperative efficiency and task success rates. Across four backbone LLMs, our method achieves average success rate improvements of 4.5% on the PARTNR-Dialog benchmark and 6.8% on the TDW-MAT benchmark over state-of-the-art open-source communicative agent frameworks. See the LLawCo project page for details: https://www.merl.com/research/highlights/LLawCo