The purpose of multi-agent systems (MAS) is to coordinate independent processes. An agent is a computer entity in the form of a program or a robot. An agent can be considered autonomous because it is able to adapt when its environment changes. A MAC consists of a set of computer processes that happen at the same time and exist at the same time, share common resources and communicate with each other. The key problem in the MAC is the formalization of coordination between agents.
Defining multi-agent systems
MAC is a forward-looking approach to software development for applications in complex domains where interacting application components are autonomous and distributed, operate in dynamic and uncertain environments, must comply with some organizational rules and laws, and can join and leave a multi-agent system during runtime.
Examples of such applications are systems thatmanage and optimize the production and distribution of electricity between consumers or systems that optimally plan loads in transport systems. The development of multi-agent systems requires the creation of separate agents, organizations and environments.
Programming languages provide programming constructs for implementing individual agents in terms of social and cognitive concepts such as information, goals, options, norms, emotions, and decision rules.
Multi-agent organizations in terms of social and organizational concepts have roles, endowed with norms, communication protocols, resources that are subject to monitoring. The developed programming languages and frameworks are used to create agent-based simulations for many industries of continuous production: electricity, metallurgy, he althcare, Internet, transportation, traffic management and serious games.
MAS differ from single-agent systems in that they have several agents that model each other's goals and actions. In a general scenario, there may be direct interaction between agents. From the point of view of a single agent, multi-agent systems differ most significantly from systems with a single agent in that the dynamics of the environment can be determined by other agents. In addition to the uncertainty that can be inherent in a domain, other agents deliberately impact the environment in unpredictable ways.
Thus, all MACs can be considered as having dynamic environments, which is typical for modernmulti-agent systems. There can be any number of agents with varying degrees of heterogeneity, with or without the possibility of direct communication.
MAS architecture
Agents must be equipped with a cognitive model:
- beliefs;
- wishes;
- intentions.
On the one hand, he reads "Beliefs" about the environment, which are the result of his knowledge and perceptions, and on the other, a set of "Desires". By crossing these two sets, a new set of "Intentions" is obtained, which are then directly translated into actions.
Agents must have a communication system. There are several specialized languages for this purpose: Language Query and Manipulation Language (KQML). Recently, the FIPA-ACL standard, created by the FIPA Foundation for Intelligent Physical Agents, has been circulated. This last principle of building multi-agent systems is based on the theory of speech acts.
The problem of adaptation is a thorny issue that is currently the subject of much research. One can give an example of some viruses, both biological and computer, able to adapt to a mutant environment.
Finally, an efficient implementation of the MAC, although not strictly speaking part of the system architecture, deserves attention in the many programming languages that have been developed for the study of artificial intelligence. In particular, the LISP language is mentioned. These architectural elements are applied to a system consisting of cognitiveagents.
Categories or models of agents
The classification of agents is based on two criteria: cognitive agents or reagents that exhibit, on the one hand, teleonomic behavior, or a reflex. The distinction that can be made between cognitive and reactive is essentially a representation of the world available to the agent. If an individual is endowed with a "symbolic representation" of the world from which he can formulate reasoning, then one speaks of a cognitive agent, whereas if he has only a "sub-symbolic representation", that is, limited to his perceptions, one speaks of a reactive agent. This cognitive and reactive distinction corresponds to two theoretical schools of multi-agent systems.
The first one supports the fundamental approach of "smart" agents for cooperation from a sociological point of view. In the second, the possibility of the emergence of "smart" behavior of a set of non-intelligent agents (ant type) is studied. The second difference between behavioral behavior and reflex separates intentional behavior, the pursuit of explicit goals, from perceptual behavior. Thus, the tendencies of agents can be explicitly expressed in agents or, conversely, come from the environment. Table grouping different types of agents:
- Cognitive agents.
- Reactive agents.
- Telenomic behavior.
- Intentional agents.
- Managed agents.
- Reflex behavior.
- Agents "modules".
- Tropical agents.
Cognitive agents are mostly intentional, i.e.they have fixed goals that they are trying to achieve. However, sometimes agents called modules are used, which have an idea of their "universe" without specific goals. They could serve, for example, to answer questions from other agents in the "universe".
Reagents can be divided into actuators and tropical agents. The instinctive agent will have a fixed mission and will trigger a behavior if it sees that the environment no longer fits its assigned purpose. Tropical agent reacts only to the local state of the environment, for example if there is light, then it runs. The source of motivation in the internal case of drive agents who have a "mission" refers only to the environment.
Organizational paradigms
With the development of such systems, different organizational paradigms have been developed. These structures of multi-agent systems set the framework for relationships and interactions between agents.
Hierarchies. In this model, agents are hierarchical according to a tree structure in which each node is an agent and has a permission link on its child nodes. This model destroys the overall purpose of the system.
Holarchy is approaching the hierarchy. There is no authority relationship between an agent and its subgroup.
A coalition is a temporary alliance of agents who come together and cooperate because their personal interests meet. The value of the coalition must be greater than the sum of the individual values of the agent components.
Congregations are very much like coalitions andcommands. However, they are meant to be permanent and usually have multiple goals to achieve. In addition, agents can enter and leave congregations and belong to several at the same time.
Society is a collection of diverse agents that interact and communicate. They have different goals, do not have the same level of rationality and the same capabilities, but they all obey common laws (norms).
Federation agents give some of their autonomy to their group's delegate. Group agents interact only with their delegate, which in turn interacts with delegates from other groups.
Sales agents offer items that buyers' agents can claim. This type of organization makes it possible, for example, to simulate real markets and compare different trading strategies.
Matrix organization agents are hierarchical. However, unlike the hierarchy presented above, where an agent is only subordinate to a few other agents, those in one matrix organization may be subject to several others.
Combinations - This combined organization mixes many of the styles above. This could be, for example, a coalition or a hierarchy of teams.
Artificial Intelligence
The goal of cognitive science is to understand the nature and workings of artificial intelligence, which is what processes internal information to make it purposeful. Many concepts fit this description: humans, computers, robots, sensory systems,the list is endless. One type of system of particular interest to cognitive scientists is the artificial self-agent that acts on information.
An intelligent agent (IA) is able to make decisions based on its experience and can choose actions in different situations. As the term "artificial" suggests, the kind of autonomous agents of interest is not something created by nature. Therefore, an artificial agent is everything that is created by people, capable of acting on the basis of the information that it perceives, its own experiences, decisions and actions.
The field of extra-natural intelligence provides the technical skills to translate desired types of agents into a programming language, related software, and appropriate architecture (hardware and related software) to implement the agent in the real or simulated world.
Environment of the world of perception
Agent is anything that takes in the environment through sensors and acts on it through effectors, which sounds simple enough. This definition of an agent covers a wide range of machines, from thermostats to objects that can actually learn a small repertoire of behavior.
Sensors are tools used by an agent to collect information about their world. The keyboard and camcorder can work as sensors if they are associated with the agent. At the end of the system's response, the performers are the tools used by the agent to influence the environment. Examples of effectors aremonitor, printer and robotic arm.
Usually the environment is the agent's domain or world. These domains, at least for now, should be limited to specific types of situations in order to avoid the unlimited possibilities of the everyday world.
Autonomous Influence System
An Autonomous Agent is “a system within and part of an environment that perceives that environment and acts on it over time in order to carry out its own agenda and in order to influence what it experiences in the future . This definition by Franklin and Greisser reflects all the basic functions of intelligent agents, except for their sociability. This provides a good approximation of the main features of the wide variety of AIs in development.
Such agents feel their environment. But here sensory data or perceptions include not only data about other objects, but also the influence of the agent itself on the state of affairs in the environment. Sensors can be organic, such as eyes and ears and their neural processors, or artificial, such as video and audio processors embedded in a digital computer. The environment can be a very limited area, like an enclosed space, or very complex, like a stock market or a collection of asteroids. Sensors must match the types of objects the agent interacts with.
Reflex type of interaction
The reflector agent has a more complex mechanism. Instead of direct dynamicin relation to the environment, he looks for what he must do in the list of rules. The reflex agent responds to a given perception with a programmed response. Even if there are thousands of possible responses to a given perception, the agent has a built-in list of situation action rules to execute those responses that have already been considered by the programmer. The situation action rule is basically a hypothetical imperative.
Reflex agents are really not very bright. They just can't handle the novelty. The intelligent agent contains the features of its less complex cousins, but is not as limited. He acts according to the agenda. It has a set of goals that it actively pursues. The target-based agent has an idea of the current state of the environment and how that environment typically works. It pursues major strategies or goals that cannot be achieved immediately. This makes the agent active, not just reactive.
Target functional utility
In more complex agents, a housekeeping measure is applied to the various possible actions that can be performed in the environment. This complex scheduler is a service-based agent. The service-based agent will evaluate each scenario to see how well it achieves certain criteria for getting a good result. Things like the likelihood of success, the resources needed to complete the scenario, the importance of the goal to be achieved, the time it would take, can all be factored into utility function calculations.
BecauseSince a programmer cannot typically predict all the states of the world an agent will encounter, the number of rules that would need to be written for a reflex agent would be astronomical even in very simple areas such as scheduling meetings or organizing transport routes and supplies.
Basic control loop
Given the definition of an intelligent agent, consider the basic control loop written by agent theorist Michael Vuladrich in 2000:
- keep peace;
- update internal world model;
- achieve a deliberate intent;
- use means/ends to get a blueprint for intentions;
- execute the plan;
- end the process.
This pattern needs some interpretation. The agent observes the world - this means that he, using his sensors, collects perceptions. The sensor may be a keyboard attached to a digital computer or a visual processor attached to a robot. It can be anything that allows the agent to gather representations of the world. Updating the internal model means that the agent adds a new perception to its sequence of perceptions and programmed information about the world.
Multi-Agent Development Platforms
AnyLogic is an open source multi-agent and multi-component CORMAS simulation software based on the object-oriented programming language SmallTalk.
DoMIS is a multi-agent systems design tool focused on "operational control of complex systems" and based on the B-ADSC design method.
JACK is a programming language and development environment for cognitive agents developed by Agent Oriented Software as an agent-oriented extension of the Java language.
GAMA is an open source modeling platform (LGPL) offering a spatially explicit agent-based modeling environment using GIS data to describe agents and their environment.
JADE (Java Agent DEVELOPMENT) is an open source multi-agent development framework based on the Java language.
Seven models of the standard
In the evolutionary process of research, there is more input on how to create a system that is reliable and represents a higher level of quality. The trend to continue is to complement or expand existing methods that have managed to consolidate decision making within development.
The methodological standard allows, in an understandable and simple way, to create a MAC, not only using natural language, but also using description templates that help in system specification.
The methodological standard offers seven models of problems or their solutions for building MAC:
- A scenario model describing a company or organization.
- The goals and objectives model defines and describes the organic structure.
- The agent model defines humans and autonomous systems.
- The role model associates goals and objectives with a particular agent.
- The organizational model describes the environment to which an individual agent is associated.
- The interaction model describes the relationship, emphasizing their coordination of agents.
- The design model defines the agent and network architecture.
Examples of interaction between agents
MAS are used to simulate the interaction between autonomous agents. The use of multi-agent systems, for example, in sociology makes it possible to parameterize the various agents that make up the community. By adding constraints, you can try to understand what will be the most effective component to achieve the expected result. They should experiment with scenarios that would not be achievable by real people, either for technical or ethical reasons.
Distributed IA was created to solve the complexity of large monolithic non-natural intelligence programs - execution, distribution and centralized control. To solve a complex problem, it is sometimes easier to create relatively small programs (agents) in cooperation than one large monolithic program. Autonomy allows the system to dynamically adapt to unforeseen changes in the environment.
Examples of multi-agent systems in the gaming industry are numerous and varied. They are used in video games and films, including the MASSIVE software, for example to simulate crowd movement in the Lord of the Rings trilogy. They may alsoused by companies, for example, to track the behavior of customers browsing websites.
MAS are also used in the world of finance. For example, the MetaTrader 4 platform allows the use of expert agents in automated trading that follow Forex rates
Benefits of using the system
In IA research, agent-based systems technology has been embraced as a new paradigm for conceptualizing, designing, and implementing software systems. Benefits of the multi-MAS approach:
- Shares computing resources and capabilities across a network of interconnected agents.
- Allows interconnection and interoperability of multiple existing legacy systems.
- Covering various fields including aircraft maintenance, book e-wallets, military mine clearance, wireless communications and communication, military logistics planning, supply chain management system, collaborative mission planning, financial portfolio management.
In research, IA technology for agent-based systems has been embraced as a new paradigm for conceptualizing, designing, implementing, and multi-agent learning of software systems.
Thus, MAC is a loosely coupled network of software agents that interact to solve problems beyond the individual ability or knowledge of each problem maker.
