Recent years have seen the application of deep reinforcement learning techniques to cooperative multi-agent systems, with great empirical success. However, given the lack of theoretical insight, it remains unclear what the employed neural networks are learning, or how we should enhance their learning power to address the problems on which they fail. In this work, we empirically investigate the learning power of various network architectures on a series of one-shot games. Despite their simplicity, these games capture many of the crucial problems that arise in the multi-agent setting, such as an exponential number of joint actions or the lack of an explicit coordination mechanism. Our results extend those in [4] and quantify how well various approaches can represent the requisite value functions, and help us identify the reasons that can impede good performance, like sparsity of the values or too tight coordination requirements.

Policy gradient methods have become one of the most popular classes of algorithms for multi-agent reinforcement learning. A key challenge, however, that is not addressed by many of these methods is multi-agent credit assignment: assessing an agent's contribution to the overall performance, which is crucial for learning good policies. We propose a novel algorithm called Dr.Reinforce that explicitly tackles this by combining difference rewards with policy gradients to allow for learning decentralized policies when the reward function is known. By differencing the reward function directly, Dr.Reinforce avoids difficulties associated with learning the Q-function as done by Counterfactual Multiagent Policy Gradients (COMA), a state-of-the-art difference rewards method. For applications where the reward function is unknown, we show the effectiveness of a version of Dr.Reinforce that learns an additional reward network that is used to estimate the difference rewards.

This paper considers optimal traffic signal control in smart cities, which has been taken as a complex networked system control problem. Given the interacting dynamics among traffic lights and road networks, attaining controller adaptivity and scalability stands out as a primary challenge. Capturing the spatial-temporal correlation among traffic lights under the framework of Multi-Agent Reinforcement Learning (MARL) is a promising solution. Nevertheless, existing MARL algorithms ignore effective information aggregation which is fundamental for improving the learning capacity of decentralized agents. In this paper, we design a new decentralized control architecture with improved environmental observability to capture the spatial-temporal correlation. Specifically, we first develop a topology-aware information aggregation strategy to extract correlation-related information from unstructured data gathered in the road network. Particularly, we transfer the road network topology into a graph shift operator by forming a diffusion process on the topology, which subsequently facilitates the construction of graph signals. A diffusion convolution module is developed, forming a new MARL algorithm, which endows agents with the capabilities of graph learning. Extensive experiments based on both synthetic and real-world datasets verify that our proposal outperforms existing decentralized algorithms.

In human societies, people often incorporate fairness in their decisions and treat reciprocally by being kind to those who act kindly. They evaluate the kindness of others' actions not only by monitoring the outcomes but also by considering the intentions. This behavioral concept can be adapted to train cooperative agents in Multi-Agent Reinforcement Learning (MARL). We propose the KindMARL method, where agents' intentions are measured by counterfactual reasoning over the environmental impact of the actions that were available to the agents. More specifically, the current environment state is compared with the estimation of the current environment state provided that the agent had chosen another action. The difference between each agent's reward, as the outcome of its action, with that of its fellow, multiplied by the intention of the fellow is then taken as the fellow's "kindness". If the result of each reward-comparison confirms the agent's superiority, it perceives the fellow's kindness and reduces its own reward. Experimental results in the Cleanup and Harvest environments show that training based on the KindMARL method enabled the agents to earn 89\% (resp. 37\%) and 44% (resp. 43\%) more total rewards than training based on the Inequity Aversion and Social Influence methods. The effectiveness of KindMARL is further supported by experiments in a traffic light control problem.

Remarkable progress has been made on automated problem solving through societies of agents based on large language models (LLMs). Existing LLM-based multi-agent systems can already solve simple dialogue tasks. Solutions to more complex tasks, however, are complicated through logic inconsistencies due to cascading hallucinations caused by naively chaining LLMs. Here we introduce MetaGPT, an innovative meta-programming framework incorporating efficient human workflows into LLM-based multi-agent collaborations. MetaGPT encodes Standardized Operating Procedures (SOPs) into prompt sequences for more streamlined workflows, thus allowing agents with human-like domain expertise to verify intermediate results and reduce errors. MetaGPT utilizes an assembly line paradigm to assign diverse roles to various agents, efficiently breaking down complex tasks into subtasks involving many agents working together. On collaborative software engineering benchmarks, MetaGPT generates more coherent solutions than previous chat-based multi-agent systems. Our project can be found at https://github.com/geekan/MetaGPT

A significant challenge facing researchers in the area of multi-agent reinforcement learning (MARL) pertains to the identification of a library that can offer fast and compatible development for multi-agent tasks and algorithm combinations, while obviating the need to consider compatibility issues. In this paper, we present MARLlib, a library designed to address the aforementioned challenge by leveraging three key mechanisms: 1) a standardized multi-agent environment wrapper, 2) an agent-level algorithm implementation, and 3) a flexible policy mapping strategy. By utilizing these mechanisms, MARLlib can effectively disentangle the intertwined nature of the multi-agent task and the learning process of the algorithm, with the ability to automatically alter the training strategy based on the current task's attributes.

One of the main challenges in offline Reinforcement Learning (RL) is the distribution shift that arises from the learned policy deviating from the data collection policy. This is often addressed by avoiding out-of-distribution (OOD) actions during policy improvement as their presence can lead to substantial performance degradation. This challenge is amplified in the offline Multi-Agent RL (MARL) setting since the joint action space grows exponentially with the number of agents. To avoid this curse of dimensionality, existing MARL methods adopt either value decomposition methods or fully decentralized training of individual agents. However, even when combined with standard conservatism principles, these methods can still result in the selection of OOD joint actions in offline MARL. To this end, we introduce AlberDICE, an offline MARL algorithm that alternatively performs centralized training of individual agents based on stationary distribution optimization. AlberDICE circumvents the exponential complexity of MARL by computing the best response of one agent at a time while effectively avoiding OOD joint action selection. Theoretically, we show that the alternating optimization procedure converges to Nash policies. In the experiments, we demonstrate that AlberDICE significantly outperforms baseline algorithms on a standard suite of MARL benchmarks.

\textit{Relative overgeneralization} (RO) occurs in cooperative multi-agent learning tasks when agents converge towards a suboptimal joint policy due to overfitting to suboptimal behavior of other agents. In early work, optimism has been shown to mitigate the \textit{RO} problem when using tabular Q-learning. However, with function approximation optimism can amplify overestimation and thus fail on complex tasks. On the other hand, recent deep multi-agent policy gradient (MAPG) methods have succeeded in many complex tasks but may fail with severe \textit{RO}. We propose a general, yet simple, framework to enable optimistic updates in MAPG methods and alleviate the RO problem. Specifically, we employ a \textit{Leaky ReLU} function where a single hyperparameter selects the degree of optimism to reshape the advantages when updating the policy. Intuitively, our method remains optimistic toward individual actions with lower returns which are potentially caused by other agents' sub-optimal behavior during learning. The optimism prevents the individual agents from quickly converging to a local optimum. We also provide a formal analysis from an operator view to understand the proposed advantage transformation. In extensive evaluations on diverse sets of tasks, including illustrative matrix games, complex \textit{Multi-agent MuJoCo} and \textit{Overcooked} benchmarks, the proposed method\footnote{Code can be found at \url{https://github.com/wenshuaizhao/optimappo}.} outperforms strong baselines on 13 out of 19 tested tasks and matches the performance on the rest.

In many real-world systems, such as adaptive robotics, achieving a single, optimised solution may be insufficient. Instead, a diverse set of high-performing solutions is often required to adapt to varying contexts and requirements. This is the realm of Quality-Diversity (QD), which aims to discover a collection of high-performing solutions, each with their own unique characteristics. QD methods have recently seen success in many domains, including robotics, where they have been used to discover damage-adaptive locomotion controllers. However, most existing work has focused on single-agent settings, despite many tasks of interest being multi-agent. To this end, we introduce Mix-ME, a novel multi-agent variant of the popular MAP-Elites algorithm that forms new solutions using a crossover-like operator by mixing together agents from different teams. We evaluate the proposed methods on a variety of partially observable continuous control tasks. Our evaluation shows that these multi-agent variants obtained by Mix-ME not only compete with single-agent baselines but also often outperform them in multi-agent settings under partial observability.

Multi-agent path finding (MAPF) is an abstract model for the navigation of multiple robots in warehouse automation, where multiple robots plan collision-free paths from the start to goal positions. Reinforcement learning (RL) has been employed to develop partially observable distributed MAPF policies that can be scaled to any number of agents. However, RL-based MAPF policies often get agents stuck in deadlock due to warehouse automation's dense and structured obstacles. This paper proposes a novel hybrid MAPF policy, RDE, based on switching among the RL-based MAPF policy, the Distance heat map (DHM)-based policy and the Escape policy. The RL-based policy is used for coordination among agents. In contrast, when no other agents are in the agent's field of view, it can get the next action by querying the DHM. The escape policy that randomly selects valid actions can help agents escape the deadlock. We conduct simulations on warehouse-like structured grid maps using state-of-the-art RL-based MAPF policies (DHC and DCC), which show that RDE can significantly improve their performance.