Heterogeneous MARL

MARL: Multi-Agent Reinforcement Learning

Overview

Heterogeneous multi-agent reinforcement learning (MARL) studies coordination when agents differ in capabilities, goals, sensors, or dynamics, as in robot teams, traffic, and network control. Methods include role-conditioned and hierarchical policies, communication and graph-attention architectures, heterogeneous-agent policy optimization such as HAPPO and HATRPO, and scalable approaches based on mean fields or federated learning. Researchers evaluate these methods in settings including Isaac Lab robotics, CybORG cyber defense, and cooperative resource-allocation tasks, often measuring coordination, transfer, fairness, and constraint satisfaction. A central challenge is balancing shared learning and efficient coordination against the need for agent-specific policies that remain robust under changing teammates and environments.

Latest papers

CardsList

35 papers7 papers in the last four weeks, up 75% on the four weeks before. 0.1% of all new papers.

  1. Learning When to Cooperate Under Heterogeneous Goals

    Mar 7, 2026Max Taylor-Davies, Neil Bramley, Christopher G. LucasMulti-Agent CollaborationAd Hoc Teamwork

  2. HALO: Learning Human-Robot Collaboration via Heterogeneous-Agent Lyapunov Policy Optimization

    Mar 4, 2026Hao Zhang, Yaru Niu, Yikai Wang +2Decentralized MARLHeterogeneous MARL

  3. Heterogeneous Agent Collaborative Reinforcement Learning

    Mar 3, 2026Zhixia Zhang, Zixuan Huang, Gongxun Li +10Multi-Agent CollaborationHeterogeneous MARL

  4. Graphon Mean-Field Subsampling for Cooperative Heterogeneous Multi-Agent Reinforcement Learning

    Feb 18, 2026Emile Anand, Richard Hoffmann, Sarah Liaw +1Reinforcement LearningCooperative MARL

  5. HARL-A: An Extensible Benchmark Framework for Heterogeneous Multi-Agent Adversarial Reinforcement Learning in IsaacLab

    Sep 26, 2025Isaac Peterson, Christopher Allred, Jacob Morrey +1RL BenchmarksAdversarial MARL