- Atlas
- Computer Science
- Multiagent Systems
Multiagent Systems
Multiagent systems is the branch of artificial intelligence that studies what happens when many autonomous programs, robots, or simulated actors share an environment and must get along.
15 frameworks, from 1971 to today · Artificial Intelligence · Computer Science
- Agent-based model 1971-Present
- Distributed Artificial Intelligence 1975-1995
- Cooperative Teamwork and Distributed Constraint Reasoning 1980-Present
- Behavior-based robotics 1986-2000
- Swarm Intelligence and Agent-Based Modeling 1987-Present
- Game-theoretic multiagent analysis 1988-Present
- Swarm intelligence 1989-Present
- Mechanism design 1990-Present
- Belief–desire–intention software model 1991-2010
- Agent Communications Language 1993-2010
- Distributed constraint reasoning 1993-Present
- Multi-agent reinforcement learning 1995-Present
- Opponent modeling 1995-Present
- Teamwork and joint intention 1996-2010
- LLM-based multiagent systems 2022-Present
Overview
Multiagent systems is the branch of artificial intelligence that studies what happens when many autonomous programs, robots, or simulated actors share an environment and must get along. An agent here means a system that perceives its world and acts on it to pursue its own goals; a multiagent system is any collection of such agents whose actions affect one another12. What is at stake is coordination without a single controller: how independent decision-makers can divide work, share information, avoid collisions and deadlock, negotiate over scarce resources, and cope with others who may be selfish, adversarial, or simply different1. The field's history is a long convergence of three streams — distributed problem-solving, the economics of strategic interaction, and the engineering of robot and software societies — that only became one recognizable discipline in the 1990s.
Distributed problem solving and the birth of DAI
The earliest strand grew out of 1970s distributed computing and speech-understanding research. The Hearsay-II speech system (Carnegie Mellon, late 1970s) split recognition among many "knowledge sources" that posted and read partial hypotheses on a shared blackboard — an early demonstration that loosely coupled specialists could solve a problem no single module could. Around the same time, work on distributed artificial intelligence (DAI) asked how a network of problem-solvers could cooperate on a task when no one node had the whole picture3. The Contract Net Protocol, proposed by Reid Smith in 1980, was the landmark: a manager announces a task, contractors bid on it, and the manager awards the contract4. It gave a concrete, reusable mechanism for task allocation by negotiation, and it framed the central DAI question — how to decompose work and share results — in terms practitioners could implement43.
Reactive robots and the behavior-based turn
A second stream came from robotics. Classical planning assumed an agent built a world model, planned, then acted. Rodney Brooks argued in the mid-1980s, most famously in "A Robust Layered Control System for a Mobile Robot" (1986), that intelligence could emerge from layered reactive behaviors — simple rules responding directly to sensor input — with no central model at all. His subsumption architecture let higher layers override lower ones, and robots built this way (the Genghis and Herbert machines of the late 1980s) walked and collected objects with remarkable robustness. The lesson for multiagent work was double-edged: simple local rules can produce coherent global behavior, but purely reactive agents struggle with anything requiring memory or long-range planning. That tension pushed the field toward hybrid designs and, later, toward learning.
Swarms, simulation, and agent-based modeling
The idea that local rules generate global order had a parallel in the study of collective behavior5. Craig Reynolds's "boids" (1987) showed that three simple steering rules — separation, alignment, cohesion — reproduced realistic flocking. In 1989 Gerardo Beni and Jing Wang coined swarm intelligence for the emergent collective behavior of decentralized robot groups, and the 1990s brought influential optimization methods modeled on insect colonies, notably Marco Dorigo's ant colony optimization (early 1990s) and Russell Eberhart and James Kennedy's particle swarm optimization (1995). Meanwhile, in the social sciences, agent-based modeling — associated with researchers such as Joshua Epstein and Robert Axtell, whose Sugarscape model appeared in the mid-1990s — used populations of simple simulated agents to study how macro patterns like segregation or wealth inequality arise from micro decisions. Thomas Schelling's earlier segregation model (1971) is the classic ancestor. These communities shared the swarm intuition but pursued different goals: optimization, control, and social explanation.
Game theory and mechanism design enter AI
Strategic interaction gave the field its mathematical backbone6. Game theory, developed by von Neumann and Morgenstern in the 1940s and extended by John Nash's equilibrium concept (1950), describes how rational agents choose when each one's payoff depends on the others67. Its central solution idea, the Nash equilibrium, is a profile of strategies from which no single agent can gain by unilaterally changing course7. From the late 1980s, AI researchers began importing these tools to analyze and design agent interactions. Mechanism design — "reverse game theory," which asks what rules produce good outcomes given that agents act in their own interest — became central to auctions and resource allocation8. The 1994 paper by Jeffrey Rosenschein and Gilad Zlotkin, Rules of Encounter, applied negotiation and game theory directly to automated agents, and the 1995 appearance of the Journal of Artificial Intelligence Research paper on the subject by Tuomas Sandholm and Victor Lesser on coalition formation signaled a maturing research program. Auction-based allocation later became a standard tool for distributed scheduling and, in the 2000s, for allocating computing resources.
BDI agents and the software-agent boom
The 1990s also saw agents move from robots and simulations into software. The belief–desire–intention (BDI) model, developed by Michael Bratman in philosophy and adapted to computing by Anand Rao and Michael Georgeff in the early 1990s, described an agent by its beliefs about the world, its desires (goals), and the intentions it has committed to. It gave programmers a principled architecture for practical reasoning, and it anchored a family of agent programming languages and platforms91011. The decade's enthusiasm for "software agents" — interface agents, information agents, mobile agents — produced standards for how agents should talk. KQML (Knowledge Query and Manipulation Language), developed in the early 1990s under Tim Finin and colleagues, and its successor FIPA-ACL, standardized by the Foundation for Intelligent Physical Agents from the late 1990s, defined message types such as inform, request, and propose, with a semantics grounded in speech-act theory. These agent communication languages let independently built agents interoperate, though the gap between formal semantics and working implementations remained a persistent complaint.
Cooperation, constraints, and teamwork
Running alongside was a body of work on how agents coordinate when they must act together. Distributed constraint reasoning — formalized in the 1990s by researchers including Makoto Yokoo and, in a widely cited 2000 survey, by Pragnesh Jay Modi and colleagues — treated coordination as a constraint-satisfaction problem spread across agents, each holding part of the problem and communicating to find a globally consistent assignment. Teamwork research, notably the STEAM system by Milind Tambe (late 1990s), addressed how agents in a team maintain shared plans and adapt when members fail. These threads established that coordination could be studied as an algorithmic problem with measurable communication costs, not just a design philosophy.
Multiagent reinforcement learning
A third major stream asked how agents learn to coordinate through experience12. Reinforcement learning — learning by trial, error, and reward — had been developed for single agents, but the multiagent case is harder because each agent's environment includes the others, who are also learning. Work in the 1990s and 2000s, associated with researchers such as Michael Littman, Caroline Claus, and Craig Boutilier, formalized multiagent RL and showed that naive independent learners can oscillate or converge to poor outcomes. The field developed techniques for joint-action learning, communication, and credit assignment, and the 2010s brought deep multiagent RL, in which neural networks approximate policies for many interacting agents. A parallel line, opponent modeling, studies how an agent can infer the strategy of others and exploit it — from simple pattern recognition in repeated games to equilibrium-based reasoning. The tension between learning a robust general policy and exploiting a specific opponent's weaknesses remains a live design choice.
The present: learning, language models, and open questions
Today the field's center of gravity has shifted toward learning-based coordination. Deep multiagent reinforcement learning underpins work on robot teams, traffic control, and game-playing, while game-theoretic mechanism design remains the toolkit for auctions and marketplaces. The most recent turn is the use of large language models as agents: since the early 2020s, systems have been built in which several LLM-driven agents converse, assign roles, and critique one another to solve tasks — a return, in new form, to the old dream of agents that coordinate through language. Whether such systems genuinely coordinate or merely produce the appearance of it is an open question, as are the field's long-standing ones: how to guarantee that self-interested agents reach socially good outcomes, how to scale coordination to thousands of heterogeneous agents, and how to verify that a multiagent system behaves safely when no one designed its global behavior. Those questions descend directly from the blackboard systems, contract nets, and boids of the field's first decades.
References
- Multi-agent system (introduction, Characteristics, Research), Wikipedia
- Artificial intelligence, Wikipedia
- Distributed artificial intelligence (Development), Wikipedia
- Contract Net Protocol, Wikipedia
- Swarm intelligence, Wikipedia
- Game theory (introduction, History), Wikipedia
- Nash equilibrium (introduction, History), Wikipedia
- Mechanism design, Wikipedia
- Belief–desire–intention software model, Wikipedia
- 3APL, Wikipedia
- AgentSpeak, Wikipedia
- Multi-agent reinforcement learning, Wikipedia
The frameworks of Multiagent Systems
Each framework is a way of working the field has used: when it held, and the question it answers.
- 1971-Present
- 1975-1995
Distributed Artificial Intelligence
How can distributed problem-solvers cooperate without a controller?
- 1980-Present
Cooperative Teamwork and Distributed Constraint Reasoning
How do cooperating agents divide and coordinate work?
- 1986-2000
Behavior-based robotics
Can intelligence emerge from layered reactive behaviors?
- 1987-Present
Swarm Intelligence and Agent-Based Modeling
How do local rules yield global order?
- 1988-Present
Game-theoretic multiagent analysis
How do self-interested agents choose when payoffs interact?
- 1989-Present
Swarm intelligence
How can colonies of simple agents solve hard problems?
- 1990-Present
Mechanism design
What rules make self-interested agents produce good outcomes?
- 1991-2010
Belief–desire–intention software model
How should an agent reason about beliefs, goals and commitments?
- 1993-2010
Agent Communications Language
How should independent agents exchange messages?
- 1993-Present
Distributed constraint reasoning
How can agents satisfy constraints held across them?
- 1995-Present
Multi-agent reinforcement learning
How do agents learn to coordinate through experience?
- 1995-Present
Opponent modeling
How can an agent infer and exploit another's strategy?
- 1996-2010
Teamwork and joint intention
How do teams maintain shared plans when members fail?
- 2022-Present
LLM-based multiagent systems
Can language-model agents genuinely coordinate through talk?
How the ideas connect
Each arrow runs from an earlier idea to the later one it shaped. The list below says what passed between them.
Cooperative Teamwork and Distributed Constraint Reasoning grew out of Distributed Artificial Intelligence
Distributed artificial intelligence produced the Contract Net Protocol and the problem of task decomposition and negotiation, which cooperative teamwork and distributed constraint reasoning took up as algorithmic problems of coordinating agents with partial information.
Carried by Reid G. Smith; Contract Net Protocol (1980).
Agent Communications Language grew out of Distributed Artificial Intelligence
Agent communication languages grew out of the DAI concern with how distributed problem-solvers exchange requests and results; KQML made the negotiation vocabulary of that tradition into a standard message language.
Carried by Tim Finin; KQML.
Distributed constraint reasoning grew out of Cooperative Teamwork and Distributed Constraint Reasoning
Distributed constraint reasoning is the formalization of the coordination problem that cooperative distributed problem solving had posed, treating coordination as constraint satisfaction spread across agents who each hold part of the problem.
Carried by Makoto Yokoo; Pragnesh Jay Modi.
Game-theoretic multiagent analysis influenced by Distributed Artificial Intelligence
Researchers in distributed AI imported game theory to analyze encounters between self-interested agents; the 1994 book Rules of Encounter applied Nash equilibria and negotiation theory directly to automated agents.
Carried by Jeffrey Rosenschein; Gilad Zlotkin; Rules of Encounter (1994).
Belief–desire–intention software model grew out of Distributed Artificial Intelligence
The software-agent branch of distributed artificial intelligence provided the setting of independently built agents that BDI architectures were designed to program, giving practical reasoning a concrete agent architecture.
Carried by Anand Rao; Michael Georgeff.
Teamwork and joint intention grew out of Cooperative Teamwork and Distributed Constraint Reasoning
Teamwork research on shared plans and joint intentions grew out of the cooperative distributed problem-solving tradition and made the maintenance of team commitments an explicit subject.
Carried by Milind Tambe; STEAM.
Swarm Intelligence and Agent-Based Modeling influenced by Behavior-based robotics
Behavior-based robotics demonstrated that simple local rules yield coherent global behavior, and the swarm-intelligence program drew on that lesson for decentralized robot groups and optimization by colonies of simple agents.
Carried by Rodney Brooks; Gerardo Beni; Jing Wang.
Multi-agent reinforcement learning influenced by Game-theoretic multiagent analysis
Game-theoretic analysis of strategic interaction supplied multagent reinforcement learning with its central difficulty, that each learner's environment contains other rational learners, and with equilibrium concepts used to judge convergence.
Carried by Michael Littman; Caroline Claus; Craig Boutilier.
Teamwork and joint intention influenced by Belief–desire–intention software model
The BDI model's notions of belief and intention were adapted to teams: joint intention research extended individual commitments to commitments that a group maintains together.
Carried by Milind Tambe.
LLM-based multiagent systems revived Agent Communications Language
The new LLM-based systems returned to the agent-communication-language ambition of independently built agents interoperating through messages, using natural language in place of standardized speech-act protocols.
Agent Communications Language influenced by Belief–desire–intention software model
BDI architectures gave agent communication languages a natural semantics for messages such as inform and request, since the mental attitudes of belief, desire and intention supplied the content that speech acts were meant to convey.
Beyond the field
Where Multiagent Systems's ideas came from in other fields, and what they led to there: the strongest 12 of 58 links.
Agent-based model influenced Agent-based macroeconomics in Macroeconomics
Agent-based macroeconomics took its simulation method directly from agent-based modeling in computer science and multiagent systems. The technique of simulating interacting rule-following agents was carried over to aggregate economic models.
Distributed Artificial Intelligence grew out of Symbolic artificial intelligence in Artificial Intelligence
Distributed artificial intelligence grew directly out of symbolic AI: early DAI work took the symbolic problem-solving agents of classical AI and distributed them across multiple communicating problem solvers.
Cooperative Teamwork and Distributed Constraint Reasoning influenced by Message-Passing Coordination Protocols in Distributed Algorithms
Distributed constraint reasoning and cooperative teamwork frameworks adopted message-passing coordination protocols from distributed algorithms, building contract-net-style and constraint-propagation schemes on that base.
Swarm Intelligence and Agent-Based Modeling influenced by Evolutionary Computation in Artificial Intelligence
Evolutionary computation supplied the population-based search and selection machinery that ant colony optimization and particle swarm optimization adapted into swarm algorithms.
Carried by Marco Dorigo; James Kennedy; Russell Eberhart.
Game-theoretic multiagent analysis led to Automated negotiation in Negotiation
Automated negotiation grew directly out of game-theoretic multiagent analysis, which modeled agents' strategic interaction and equilibrium outcomes. Negotiation agents adopted its solution concepts, utility models and equilibrium analysis as their core framework.
Game-theoretic multiagent analysis grew out of Game Theory in Applied Mathematics
Game-theoretic multiagent analysis is a direct application of game theory to settings with several computational agents, extending equilibrium and solution concepts to artificial agent populations.
Carried by John von Neumann; Oskar Morgenstern; Theory of Games and Economic Behavior (1944).
Game-theoretic multiagent analysis influenced Multi-Agent Reinforcement Learning in Reinforcement Learning
Game-theoretic multiagent analysis supplied stochastic games, Nash equilibria and convergence criteria that multi-agent RL adopted as its formal setting, notably in Littman's Markov games work.
Carried by Michael L. Littman.
Multi-agent reinforcement learning grew out of Reinforcement learning in Artificial Intelligence
Multi-agent reinforcement learning grew directly out of single-agent reinforcement learning by extending its value-based and policy-learning methods to settings with multiple interacting learners. The multi-agent framework is a generalization of the parent theory.
Multi-agent reinforcement learning led to Learning-Based Coordination in Multi Robot Systems
Learning-based coordination of robot teams is largely multi-agent reinforcement learning applied to physical systems: agents learn decentralized policies through shared reward and centralized training. The MARL formulations of joint action learning and credit assignment are the direct source of the later methods.
Agent-based model influenced Whole-Cell and Agent-Based Modelling in Systems Biology
The multiagent-systems tradition supplied the formal agent concept and simulation architectures that whole-cell modellers adapted, treating molecular components as agents with rules. This computational lineage underlies agent-based cell simulation.
Agent-based model influenced by System dynamics in Systems Simulation
System dynamics, developed by Jay Forrester at MIT in the 1950s and 1960s, supplied the stock-and-flow, feedback-loop simulation style that early agent-based modelling drew on and reacted against by adding heterogeneous discrete agents.
Carried by Jay Forrester; Industrial Dynamics (1961).
Agent-based model influenced Whole-Cell Modelling in Systems Biology
The multiagent-systems tradition supplied the formal agent concept and simulation platforms that whole-cell modellers adapted to represent molecular components. This computational lineage underlies agent-based whole-cell simulation.

Multiagent Systems is a planet in Nousfarer
Its city has a tower for each of these 15 frameworks. Land, walk in, and climb one concept at a time.
Land on Multiagent SystemsRelated fields
Fields its ideas are linked to
- Multi Robot Systems 11 links
- Artificial Intelligence 6 links
- Systems Simulation 6 links
- Ai Safety 4 links
- Reinforcement Learning 3 links
- Negotiation 3 links
- Systems Biology 2 links
- Neoclassical Economics 2 links
- Distributed Algorithms 1 link
- Game Theory 1 link
- Behavioral Game Theory 1 link
- Poker Strategy 1 link
More in Artificial Intelligence
- Ai Ethics And Fairness
- Ai Safety
- Artificial Intelligence
- Computer Vision
- Deep Learning
- Evolutionary Computation
- Knowledge Representation Reasoning
- Machine Learning
- Natural Language Processing
- Planning And Search
- Probabilistic Ai
- Reinforcement Learning
- Speech Recognition And Synthesis
See where Multiagent Systems sits on the Map of Human Knowledge