This paper introduces LLM-MARL, a unified framework that incorporates large language models (LLMs) into multi-agent reinforcement learning (MARL) to enhance coordination, communication, and generalization in simulated game environments. The framework features three modular components of Coordinator, Communicator, and Memory, which dynamically generate subgoals, facilitate symbolic inter-agent messaging, and support episodic recall. Training combines PPO with a language-conditioned loss and LLM query gating. LLM-MARL is evaluated in Google Research Football, MAgent Battle, and StarCraft II. Results show consistent improvements over MAPPO and QMIX in win rate, coordination score, and zero-shot generalization. Ablation studies demonstrate that subgoal generation and language-based messaging each contribute significantly to performance gains. Qualitative analysis reveals emergent behaviors such as role specialization and communication-driven tactics. By bridging language modeling and policy learning, this work contributes to the design of intelligent, cooperative agents in interactive simulations. It offers a path forward for leveraging LLMs in multi-agent systems used for training, games, and human-AI collaboration.
As artificial intelligence surpasses human performance in select tasks, disclosing superhuman capabilities poses distinct challenges for fairness, accountability, and trust. However, the impact of such disclosures on diverse user attitudes and behaviors remains unclear, particularly concerning potential negative reactions like discouragement or overreliance. This paper investigates these effects by utilizing Persona Cards: a validated, standardized set of synthetic personas designed to simulate diverse user reactions and fairness perspectives. We conducted an ethics board-approved study (N=32), utilizing these personas to investigate how capability disclosure influenced behaviors with a superhuman game AI in competitive StarCraft II scenarios. Our results reveal transparency is double-edged: while disclosure could alleviate suspicion, it also provoked frustration and strategic defeatism among novices in cooperative scenarios, as well as overreliance in competitive contexts. Experienced and competitive players interpreted disclosure as confirmation of an unbeatable opponent, shifting to suboptimal goals. We release the Persona Cards Dataset, including profiles, prompts, interaction logs, and protocols, to foster reproducible research into human alignment AI design. This work demonstrates that transparency is not a cure-all; successfully leveraging disclosure to enhance trust and accountability requires careful tailoring to user characteristics, domain norms, and specific fairness objectives.
We introduce Attentive VLM Agent (AVA), a multimodal StarCraft II agent that aligns artificial agent perception with the human gameplay experience. Traditional frameworks such as SMAC rely on abstract state representations that diverge significantly from human perception, limiting the ecological validity of agent behavior. Our agent addresses this limitation by incorporating RGB visual inputs and natural language observations that more closely simulate human cognitive processes during gameplay. The AVA architecture consists of three integrated components: (1) a vision-language model enhanced with specialized self-attention mechanisms for strategic unit targeting and battlefield assessment, (2) a retrieval-augmented generation system that leverages domain-specific StarCraft II knowledge to inform tactical decisions, and (3) a dynamic role-based task distribution system that enables coordinated multi-agent behavior. The experimental evaluation in our proposed AVACraft environment, which contains 21 multimodal StarCraft II scenarios, demonstrates that AVA powered by foundation models (specifically Qwen-VL and GPT-4o) can execute complex tactical maneuvers without explicit training, achieving comparable performance to traditional MARL methods that require substantial training iterations. This work establishes a foundation for developing human-aligned StarCraft II agents and advances the broader research agenda of multimodal game AI. Our implementation is available at https://github.com/camel-ai/VLM-Play-StarCraft2.
In the context of multi-agent reinforcement learning, generalization is a challenge to solve various tasks that may require different joint policies or coordination without relying on policies specialized for each task. We refer to this type of problem as a multi-task, and we train agents to be versatile in this multi-task setting through a single training process. To address this challenge, we introduce TRajectory-class-Aware Multi-Agent reinforcement learning (TRAMA). In TRAMA, agents recognize a task type by identifying the class of trajectories they are experiencing through partial observations, and the agents use this trajectory awareness or prediction as additional information for action policy. To this end, we introduce three primary objectives in TRAMA: (a) constructing a quantized latent space to generate trajectory embeddings that reflect key similarities among them; (b) conducting trajectory clustering using these trajectory embeddings; and (c) building a trajectory-class-aware policy. Specifically for (c), we introduce a trajectory-class predictor that performs agent-wise predictions on the trajectory class; and we design a trajectory-class representation model for each trajectory class. Each agent takes actions based on this trajectory-class representation along with its partial observation for task-aware execution. The proposed method is evaluated on various tasks, including multi-task problems built upon StarCraft II. Empirical results show further performance improvements over state-of-the-art baselines.
Multi-agent reinforcement learning (MARL) faces challenges in coordinating agents due to complex interdependencies within multi-agent systems. Most MARL algorithms use the simultaneous decision-making paradigm but ignore the action-level dependencies among agents, which reduces coordination efficiency. In contrast, the sequential decision-making paradigm provides finer-grained supervision for agent decision order, presenting the potential for handling dependencies via better decision order management. However, determining the optimal decision order remains a challenge. In this paper, we introduce Action Generation with Plackett-Luce Sampling (AGPS), a novel mechanism for agent decision order optimization. We model the order determination task as a Plackett-Luce sampling process to address issues such as ranking instability and vanishing gradient during the network training process. AGPS realizes credit-based decision order determination by establishing a bridge between the significance of agents' local observations and their decision credits, thus facilitating order optimization and dependency management. Integrating AGPS with the Multi-Agent Transformer, we propose the Prioritized Multi-Agent Transformer (PMAT), a sequential decision-making MARL algorithm with decision order optimization. Experiments on benchmarks including StarCraft II Multi-Agent Challenge, Google Research Football, and Multi-Agent MuJoCo show that PMAT outperforms state-of-the-art algorithms, greatly enhancing coordination efficiency.
StarCraft II is a complex and dynamic real-time strategy (RTS) game environment, which is very suitable for artificial intelligence and reinforcement learning research. To address the problem of Large Language Model(LLM) learning in complex environments through self-reflection, we propose a Reflection of Episodes(ROE) framework based on expert experience and self-experience. This framework first obtains key information in the game through a keyframe selection method, then makes decisions based on expert experience and self-experience. After a game is completed, it reflects on the previous experience to obtain new self-experience. Finally, in the experiment, our method beat the robot under the Very Hard difficulty in TextStarCraft II. We analyze the data of the LLM in the process of the game in detail, verified its effectiveness.
Since the emergence of the Large Language Model (LLM), LLM has been widely used in fields such as writing, translating, and searching. However, there is still great potential for LLM-based methods in handling complex tasks such as decision-making in the StarCraft II environment. To address problems such as lack of relevant knowledge and poor control over subtasks of varying importance, we propose a Hierarchical Expert Prompt (HEP) for LLM. Our method improves the understanding of game situations through expert-level tactical knowledge, improving the processing quality of tasks of varying importance through a hierarchical framework. Our approach defeated the highest level (Elite) standard built-in agent in TextStarCraft II for the first time and consistently outperformed the baseline method in other difficulties. Our experiments suggest that the proposed method is a practical solution for tackling complex decision-making challenges. The replay video can be viewed on https://www.bilibili.com/video/BV1uz42187EF and https://youtu.be/dO3PshWLV5M, and our codes have been open-sourced on https://github.com/luchang1113/HEP-LLM-play-StarCraftII.
Multi-Agent Reinforcement Learning (MARL) has gained significant traction for solving complex real-world tasks, but the inherent stochasticity and uncertainty in these environments pose substantial challenges to efficient and robust policy learning. While Distributional Reinforcement Learning has been successfully applied in single-agent settings to address risk and uncertainty, its application in MARL is substantially limited. In this work, we propose a novel approach that integrates distributional learning with a safety-focused loss function to improve convergence in cooperative MARL tasks. Specifically, we introduce a Barrier Function based loss that leverages safety metrics, identified from inherent faults in the system, into the policy learning process. This additional loss term helps mitigate risks and encourages safer exploration during the early stages of training. We evaluate our method in the StarCraft II micromanagement benchmark, where our approach demonstrates improved convergence and outperforms state-of-the-art baselines in terms of both safety and task completion. Our results suggest that incorporating safety considerations can significantly enhance learning performance in complex, multi-agent environments.
Artificial intelligence (AI) has enabled agents to master complex video games, from first-person shooters like Counter-Strike to real-time strategy games such as StarCraft II and racing games like Gran Turismo. While these achievements are notable, applying these AI methods in commercial video game production remains challenging due to computational constraints. In commercial scenarios, the majority of computational resources are allocated to 3D rendering, leaving limited capacity for AI methods, which often demand high computational power, particularly those relying on pixel-based sensors. Moreover, the gaming industry prioritizes creating human-like behavior in AI agents to enhance player experience, unlike academic models that focus on maximizing game performance. This paper introduces a novel methodology for training neural networks via imitation learning to play a complex, commercial-standard, VALORANT-like 2v2 tactical shooter game, requiring only modest CPU hardware during inference. Our approach leverages an innovative, pixel-free perception architecture using a small set of ray-cast sensors, which capture essential spatial information efficiently. These sensors allow AI to perform competently without the computational overhead of traditional methods. Models are trained to mimic human behavior using supervised learning on human trajectory data, resulting in realistic and engaging AI agents. Human evaluation tests confirm that our AI agents provide human-like gameplay experiences while operating efficiently under computational constraints. This offers a significant advancement in AI model development for tactical shooter games and possibly other genres.
Recently, deep Multi-Agent Reinforcement Learning (MARL) has demonstrated its potential to tackle complex cooperative tasks, pushing the boundaries of AI in collaborative environments. However, the efficiency of these systems is often compromised by inadequate sample utilization and a lack of diversity in learning strategies. To enhance MARL performance, we introduce a novel sample reuse approach that dynamically adjusts policy updates based on observation novelty. Specifically, we employ a Random Network Distillation (RND) network to gauge the novelty of each agent's current state, assigning additional sample update opportunities based on the uniqueness of the data. We name our method Multi-Agent Novelty-GuidEd sample Reuse (MANGER). This method increases sample efficiency and promotes exploration and diverse agent behaviors. Our evaluations confirm substantial improvements in MARL effectiveness in complex cooperative scenarios such as Google Research Football and super-hard StarCraft II micromanagement tasks.
This paper introduces a new environment LLM-PySC2 (the Large Language Model StarCraft II Learning Environment), a platform derived from DeepMind's StarCraft II Learning Environment that serves to develop Large Language Models (LLMs) based decision-making methodologies. This environment is the first to offer the complete StarCraft II action space, multi-modal observation interfaces, and a structured game knowledge database, which are seamlessly connected with various LLMs to facilitate the research of LLMs-based decision-making. To further support multi-agent research, we developed an LLM collaborative framework that supports multi-agent concurrent queries and multi-agent communication. In our experiments, the LLM-PySC2 environment is adapted to be compatible with the StarCraft Multi-Agent Challenge (SMAC) task group and provided eight new scenarios focused on macro-decision abilities. We evaluated nine mainstream LLMs in the experiments, and results show that sufficient parameters are necessary for LLMs to make decisions, but improving reasoning ability does not directly lead to better decision-making outcomes. Our findings further indicate the importance of enabling large models to learn autonomously in the deployment environment through parameter training or train-free learning techniques. Ultimately, we expect that the LLM-PySC2 environment can promote research on learning methods for LLMs, helping LLM-based methods better adapt to task scenarios.
Various works have aimed at combining the inference efficiency of recurrent models and training parallelism of multi-head attention for sequence modeling. However, most of these works focus on tasks with fixed-dimension observation spaces, such as individual tokens in language modeling or pixels in image completion. To handle an observation space of varying size, we propose a novel algorithm that alternates between cross-attention between a 2D latent state and observation, and a discounted cumulative sum over the sequence dimension to efficiently accumulate historical information. We find this resampling cycle is critical for performance. To evaluate efficient sequence modeling in this domain, we introduce two multi-agent intention tasks: simulated agents chasing bouncing particles and micromanagement analysis in professional StarCraft II games. Our algorithm achieves comparable accuracy with a lower parameter count, faster training and inference compared to existing methods.
Creation and storage of datasets are often overlooked input costs in machine learning, as many datasets are simple image label pairs or plain text. However, datasets with more complex structures, such as those from the real time strategy game StarCraft II, require more deliberate thought and strategy to reduce cost of ownership. We introduce a serialization framework for StarCraft II that reduces the cost of dataset creation and storage, as well as improving usage ergonomics. We benchmark against the most comparable existing dataset from \textit{AlphaStar-Unplugged} and highlight the benefit of our framework in terms of both the cost of creation and storage. We use our dataset to train deep learning models that exceed the performance of comparable models trained on other datasets. The dataset conversion and usage framework introduced is open source and can be used as a framework for datasets with similar characteristics such as digital twin simulations. Pre-converted StarCraft II tournament data is also available online.
Offline reinforcement learning (RL) has garnered significant attention for its ability to learn effective policies from pre-collected datasets without the need for further environmental interactions. While promising results have been demonstrated in single-agent settings, offline multi-agent reinforcement learning (MARL) presents additional challenges due to the large joint state-action space and the complexity of multi-agent behaviors. A key issue in offline RL is the distributional shift, which arises when the target policy being optimized deviates from the behavior policy that generated the data. This problem is exacerbated in MARL due to the interdependence between agents' local policies and the expansive joint state-action space. Prior approaches have primarily addressed this challenge by incorporating regularization in the space of either Q-functions or policies. In this work, we introduce a regularizer in the space of stationary distributions to better handle distributional shift. Our algorithm, ComaDICE, offers a principled framework for offline cooperative MARL by incorporating stationary distribution regularization for the global learning policy, complemented by a carefully structured multi-agent value decomposition strategy to facilitate multi-agent training. Through extensive experiments on the multi-agent MuJoCo and StarCraft II benchmarks, we demonstrate that ComaDICE achieves superior performance compared to state-of-the-art offline MARL methods across nearly all tasks.
Value factorization is a popular paradigm for designing scalable multi-agent reinforcement learning algorithms. However, current factorization methods make choices without full justification that may limit their performance. For example, the theory in prior work uses stateless (i.e., history) functions, while the practical implementations use state information -- making the motivating theory a mismatch for the implementation. Also, methods have built off of previous approaches, inheriting their architectures without exploring other, potentially better ones. To address these concerns, we formally analyze the theory of using the state instead of the history in current methods -- reconnecting theory and practice. We then introduce DuelMIX, a factorization algorithm that learns distinct per-agent utility estimators to improve performance and achieve full expressiveness. Experiments on StarCraft II micromanagement and Box Pushing tasks demonstrate the benefits of our intuitions.
Multi-agent proximal policy optimization (MAPPO) has recently demonstrated state-of-the-art performance on challenging multi-agent reinforcement learning tasks. However, MAPPO still struggles with the credit assignment problem, wherein the sheer difficulty in ascribing credit to individual agents' actions scales poorly with team size. In this paper, we propose a multi-agent reinforcement learning algorithm that adapts recent developments in credit assignment to improve upon MAPPO. Our approach leverages partial reward decoupling (PRD), which uses a learned attention mechanism to estimate which of a particular agent's teammates are relevant to its learning updates. We use this estimate to dynamically decompose large groups of agents into smaller, more manageable subgroups. We empirically demonstrate that our approach, PRD-MAPPO, decouples agents from teammates that do not influence their expected future reward, thereby streamlining credit assignment. We additionally show that PRD-MAPPO yields significantly higher data efficiency and asymptotic performance compared to both MAPPO and other state-of-the-art methods across several multi-agent tasks, including StarCraft II. Finally, we propose a version of PRD-MAPPO that is applicable to \textit{shared} reward settings, where PRD was previously not applicable, and empirically show that this also leads to performance improvements over MAPPO.
Value function factorization methods are commonly used in cooperative multi-agent reinforcement learning, with QMIX receiving significant attention. Many QMIX-based methods introduce monotonicity constraints between the joint action value and individual action values to achieve decentralized execution. However, such constraints limit the representation capacity of value factorization, restricting the joint action values it can represent and hindering the learning of the optimal policy. To address this challenge, we propose the Potentially Optimal Joint Actions Weighted QMIX (POWQMIX) algorithm, which recognizes the potentially optimal joint actions and assigns higher weights to the corresponding losses of these joint actions during training. We theoretically prove that with such a weighted training approach the optimal policy is guaranteed to be recovered. Experiments in matrix games, difficulty-enhanced predator-prey, and StarCraft II Multi-Agent Challenge environments demonstrate that our algorithm outperforms the state-of-the-art value-based multi-agent reinforcement learning methods.
Cooperative Multi-Agent Reinforcement Learning (MARL) necessitates seamless collaboration among agents, often represented by an underlying relation graph. Existing methods for learning this graph primarily focus on agent-pair relations, neglecting higher-order relationships. While several approaches attempt to extend cooperation modelling to encompass behaviour similarities within groups, they commonly fall short in concurrently learning the latent graph, thereby constraining the information exchange among partially observed agents. To overcome these limitations, we present a novel approach to infer the Group-Aware Coordination Graph (GACG), which is designed to capture both the cooperation between agent pairs based on current observations and group-level dependencies from behaviour patterns observed across trajectories. This graph is further used in graph convolution for information exchange between agents during decision-making. To further ensure behavioural consistency among agents within the same group, we introduce a group distance loss, which promotes group cohesion and encourages specialization between groups. Our evaluations, conducted on StarCraft II micromanagement tasks, demonstrate GACG's superior performance. An ablation study further provides experimental evidence of the effectiveness of each component of our method.
Current approaches to learning cooperative multi-agent behaviors assume relatively restrictive settings. In standard fully cooperative multi-agent reinforcement learning, the learning algorithm controls $\textit{all}$ agents in the scenario, while in ad hoc teamwork, the learning algorithm usually assumes control over only a $\textit{single}$ agent in the scenario. However, many cooperative settings in the real world are much less restrictive. For example, in an autonomous driving scenario, a company might train its cars with the same learning algorithm, yet once on the road, these cars must cooperate with cars from another company. Towards expanding the class of scenarios that cooperative learning methods may optimally address, we introduce $N$-agent ad hoc teamwork (NAHT), where a set of autonomous agents must interact and cooperate with dynamically varying numbers and types of teammates. This paper formalizes the problem, and proposes the Policy Optimization with Agent Modelling (POAM) algorithm. POAM is a policy gradient, multi-agent reinforcement learning approach to the NAHT problem, that enables adaptation to diverse teammate behaviors by learning representations of teammate behaviors. Empirical evaluation on tasks from the multi-agent particle environment and StarCraft II shows that POAM improves cooperative task returns compared to baseline approaches, and enables out-of-distribution generalization to unseen teammates.
Effective agent coordination is crucial in cooperative Multi-Agent Reinforcement Learning (MARL). While agent cooperation can be represented by graph structures, prevailing graph learning methods in MARL are limited. They rely solely on one-step observations, neglecting crucial historical experiences, leading to deficient graphs that foster redundant or detrimental information exchanges. Additionally, high computational demands for action-pair calculations in dense graphs impede scalability. To address these challenges, we propose inferring a Latent Temporal Sparse Coordination Graph (LTS-CG) for MARL. The LTS-CG leverages agents' historical observations to calculate an agent-pair probability matrix, where a sparse graph is sampled from and used for knowledge exchange between agents, thereby simultaneously capturing agent dependencies and relation uncertainty. The computational complexity of this procedure is only related to the number of agents. This graph learning process is further augmented by two innovative characteristics: Predict-Future, which enables agents to foresee upcoming observations, and Infer-Present, ensuring a thorough grasp of the environmental context from limited data. These features allow LTS-CG to construct temporal graphs from historical and real-time information, promoting knowledge exchange during policy learning and effective collaboration. Graph learning and agent training occur simultaneously in an end-to-end manner. Our demonstrated results on the StarCraft II benchmark underscore LTS-CG's superior performance.