Intelligent Agent and Multi-Agent System

A special issue of Information (ISSN 2078-2489). This special issue belongs to the section "Artificial Intelligence".

Deadline for manuscript submissions: 31 May 2025 | Viewed by 15807

Special Issue Editor


E-Mail Website
Guest Editor
Institute of Engineering, Tokyo University of Agriculture and Technology, Tokyo 184-8588, Japan
Interests: multiagent system; natural language processing; text mining; artificial intelligence

Special Issue Information

Dear Colleagues,

Research conducted on intelligent agents and multiagent systems has matured over the last decade, with many effective applications of this technology having been deployed. Despite the fact that artificial intelligence (AI) technologies for multiagent systems have only mainly emerged in recent decades, a variety of approaches has been proposed for solving certain real-life problems. It has allowed for the development of distributed and intelligent applications in complex and dynamic environments, also playing a crucial role in life, evidenced by the broad range of areas of application involved in their use, including manufacturing, management sciences, e-commerce, etc.

The main aim of this Special Issue is to seek high-quality submissions focusing on theoretical and practical aspects of multiagent systems, including theories, applications and simulations concerning AI technology. These include, but are not limited to, concepts, systems and applications of agent technologies.

Dr. Katsuhide Fujita
Guest Editor

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Information is an international peer-reviewed open access monthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 1600 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

Keywords

  • coordination, organizations, institutions and norms for multiagent systems
  • markets, auctions and noncooperative game theory
  • social choice and cooperative game theory
  • knowledge representation, reasoning and planning
  • learning and adaptation
  • modelling and simulation of societies
  • humans and AI/human–agent interaction
  • engineering multiagent systems
  • robotics
  • innovative applications of multiagent systems

Published Papers (9 papers)

Order results
Result details
Select all
Export citation of selected articles as:

Research

Jump to: Review

23 pages, 1687 KiB  
Article
Production Scheduling Based on a Multi-Agent System and Digital Twin: A Bicycle Industry Case
by Vasilis Siatras, Emmanouil Bakopoulos, Panagiotis Mavrothalassitis, Nikolaos Nikolakis and Kosmas Alexopoulos
Information 2024, 15(6), 337; https://doi.org/10.3390/info15060337 - 6 Jun 2024
Viewed by 307
Abstract
The emerging digitalization in today’s industrial environments allows manufacturers to store online knowledge about production and use it to make better informed management decisions. This paper proposes a multi-agent framework enhanced with digital twin (DT) for production scheduling and optimization. Decentralized scheduling agents [...] Read more.
The emerging digitalization in today’s industrial environments allows manufacturers to store online knowledge about production and use it to make better informed management decisions. This paper proposes a multi-agent framework enhanced with digital twin (DT) for production scheduling and optimization. Decentralized scheduling agents interact to efficiently manage the work allocation in different segments of production. A DT is used to evaluate the performance of different scheduling decisions and to avoid potential risks and bottlenecks. Production managers can supervise the system’s decision-making processes and manually regulate them online. The multi-agent system (MAS) uses asset administration shells (AASs) for data modelling and communication, enabling interoperability and scalability. The framework was deployed and tested in an industrial pilot coming from the bicycle production industry, optimizing and controlling the short-term production schedule of the different departments. The evaluation resulted in a higher production rate, thus achieving higher production volume in a shorter time span. Managers were also able to coordinate schedules from different departments in a dynamic way and achieve early bottleneck detection. Full article
(This article belongs to the Special Issue Intelligent Agent and Multi-Agent System)
19 pages, 2056 KiB  
Article
Locally Centralized Execution for Less Redundant Computation in Multi-Agent Cooperation
by Yidong Bai and Toshiharu Sugawara
Information 2024, 15(5), 279; https://doi.org/10.3390/info15050279 - 14 May 2024
Viewed by 631
Abstract
Decentralized execution is a widely used framework in multi-agent reinforcement learning. However, it has a well-known but neglected shortcoming, redundant computation, that is, the same/similar computation is performed redundantly in different agents owing to their overlapping observations. This study proposes a novel method, [...] Read more.
Decentralized execution is a widely used framework in multi-agent reinforcement learning. However, it has a well-known but neglected shortcoming, redundant computation, that is, the same/similar computation is performed redundantly in different agents owing to their overlapping observations. This study proposes a novel method, the locally centralized team transformer (LCTT), to address this problem. This method first proposes a locally centralized execution framework that autonomously determines some agents as leaders that generate instructions and other agents as workers to act according to the received instructions without running their policy networks. For the LCTT, we subsequently propose the team-transformer (T-Trans) structure, which enables leaders to generate targeted instructions for each worker, and the leadership shift, which enables agents to determine those that should instruct or be instructed by others. The experimental results demonstrated that the proposed method significantly reduces redundant computations without decreasing rewards and achieves faster learning convergence. Full article
(This article belongs to the Special Issue Intelligent Agent and Multi-Agent System)
Show Figures

Figure 1

17 pages, 2692 KiB  
Article
Proactive Agent Behaviour in Dynamic Distributed Constraint Optimisation Problems
by Brighter Agyemang, Fenghui Ren and Jun Yan
Information 2024, 15(5), 255; https://doi.org/10.3390/info15050255 - 2 May 2024
Viewed by 771
Abstract
In multi-agent systems, the Dynamic Distributed Constraint Optimisation Problem (D-DCOP) framework is pivotal, allowing for the decomposition of global objectives into agent constraints. Proactive agent behaviour is crucial in such systems, enabling agents to anticipate future changes and adapt accordingly. Existing approaches, like [...] Read more.
In multi-agent systems, the Dynamic Distributed Constraint Optimisation Problem (D-DCOP) framework is pivotal, allowing for the decomposition of global objectives into agent constraints. Proactive agent behaviour is crucial in such systems, enabling agents to anticipate future changes and adapt accordingly. Existing approaches, like Proactive Dynamic DCOP (PD-DCOP) algorithms, often necessitate a predefined environment model. We address the problem of enabling proactive agent behaviour in D-DCOPs where the dynamics model of the environment is unknown. Specifically, we propose an approach where agents learn local autoregressive models from observations, predicting future states to inform decision-making. To achieve this, we present a temporal experience-sharing message-passing algorithm that leverages dynamic agent connections and a distance metric to collate training data. Our approach outperformed baseline methods in a search-and-extinguish task using the RoboCup Rescue Simulator, achieving better total building damage. The experimental results align with prior work on the significance of decision-switching costs and demonstrate improved performance when the switching cost is combined with a learned model. Full article
(This article belongs to the Special Issue Intelligent Agent and Multi-Agent System)
Show Figures

Figure 1

17 pages, 509 KiB  
Article
Snapshot-Optimal Real-Time Ride Sharing
by Afzaal Hassan, Mark Wallace, Irene Moser and Daniel D. Harabor
Information 2024, 15(4), 174; https://doi.org/10.3390/info15040174 - 22 Mar 2024
Viewed by 836
Abstract
Ridesharing effectively tackles urban mobility challenges by providing a service comparable to private vehicles while minimising resource usage. Our research primarily concentrates on dynamic ridesharing, which conventionally involves connecting drivers with passengers in need of transportation. The process of one-to-one matching presents a [...] Read more.
Ridesharing effectively tackles urban mobility challenges by providing a service comparable to private vehicles while minimising resource usage. Our research primarily concentrates on dynamic ridesharing, which conventionally involves connecting drivers with passengers in need of transportation. The process of one-to-one matching presents a complex challenge, particularly when addressing it on a large scale, as the substantial number of potential matches make the attainment of a global optimum a challenging endeavour. This paper aims to address the absence of an optimal approach for dynamic ridesharing by refraining from the conventional heuristic-based methods commonly used to achieve timely solutions in large-scale ride-matching. Instead, we propose a novel approach that provides snapshot-optimal solutions for various forms of one-to-one matching while ensuring they are generated within an acceptable timeframe for service providers. Additionally, we introduce and solve a new variant in which the system itself provides the vehicles. The efficacy of our methodology is substantiated through experiments carried out with real-world data extracted from the openly available New York City taxicab dataset. Full article
(This article belongs to the Special Issue Intelligent Agent and Multi-Agent System)
Show Figures

Figure 1

17 pages, 2117 KiB  
Article
Online Planning for Autonomous Mobile Robots with Different Objectives in Warehouse Commissioning Task
by Satoshi Warita and Katsuhide Fujita
Information 2024, 15(3), 130; https://doi.org/10.3390/info15030130 - 26 Feb 2024
Viewed by 1031
Abstract
Recently, multi-agent systems have become widespread as essential technologies for various practical problems. An essential problem in multi-agent systems is collaborative automating picking and delivery operations in warehouses. The warehouse commissioning task involves finding specified items in a warehouse and moving them to [...] Read more.
Recently, multi-agent systems have become widespread as essential technologies for various practical problems. An essential problem in multi-agent systems is collaborative automating picking and delivery operations in warehouses. The warehouse commissioning task involves finding specified items in a warehouse and moving them to a specified location using robots. This task is defined as a spatial task-allocation problem (SPATAP) based on a Markov decision process (MDP). It is considered a decentralized multi-agent system rather than a system that manages and optimizes agents in a central manner. Existing research on SPATAP involving modeling the environment as a MDP and applying Monte Carlo tree searches has shown that this approach is efficient. However, there has not been sufficient research into scenarios in which all agents are provided a common plan despite the fact that their actions are decided independently. Thus, previous studies have not considered cooperative robot behaviors with different goals, and the problem where each robot has different goals has not been studied extensively. In terms of the cooperative element, the item exchange approach has not been considered effectively in previous studies. Therefore, in this paper, we focus on the problem of each robot being assigned a different task to optimize the percentage of picking and delivering items in time in social situations. We propose an action-planning method based on the Monte Carlo tree search and an item-exchange method between agents. We also generate a simulator to evaluate the proposed methods. The results of simulations demonstrate that the achievement rate is improved in small- and medium-sized warehouses. However, the achievement rate did not improve in large warehouses because the average distance from the depot to the items increased. Full article
(This article belongs to the Special Issue Intelligent Agent and Multi-Agent System)
Show Figures

Figure 1

21 pages, 13910 KiB  
Article
An Agent-Based Simulation Platform for a Safe Election: From Design to Simulation
by Ali V. Barenji, Benoit Montreuil, Sevda Babalou, Dima Nazzal, Frederick Benaben and Richard DeMillo
Information 2023, 14(10), 529; https://doi.org/10.3390/info14100529 - 28 Sep 2023
Viewed by 1750
Abstract
Managing the logistics and safety of an election system, from delivering voting machines to the right locations at the right time to ensuring that voting lines remain reasonable in length is a complex problem due to the scarcity of resources, especially human poll [...] Read more.
Managing the logistics and safety of an election system, from delivering voting machines to the right locations at the right time to ensuring that voting lines remain reasonable in length is a complex problem due to the scarcity of resources, especially human poll workers, and the impact of human behavior and disrupting events on the performance of this critical system. These complexities grew with the need for physical distancing during the COVID-19 pandemic coinciding with multiple key national elections, including the 2020 general presidential election in the USA. In this paper, we propose a digital clone platform leveraging agent-based simulation to model and experiment with resource allocation decisions and voter turnout fluctuations and facilitate “what-if” scenario testing of any election. As a use case, we consider three different concurrent polling location problems, namely, resource allocation, polling layout, and management. The main aim is to reduce voter waiting time and provide visibility of different scenarios for polling and state-level managers. We explain the proposed simulation platform based on Fulton County for the 2020 presidential US election. Fulton County had 238 polling locations in 2020, which provided publicly available voter turnout data. The developed platform realistically models at the county level and at specific locations, suggesting the possible allocation of finite resources among locations in the county and the configuration of each location, accounting for physical, legal, and technical constraints. Multiple realistic scenarios were developed and embedded into the simulation platform to evaluate and verify the different systems. The system performance and key attributes of the election system, such as waiting time, resource utilization, and layout safety, were tested and validated. Full article
(This article belongs to the Special Issue Intelligent Agent and Multi-Agent System)
Show Figures

Figure 1

20 pages, 8979 KiB  
Article
Modeling Theory of Mind in Dyadic Games Using Adaptive Feedback Control
by Ismael T. Freire, Xerxes D. Arsiwalla, Jordi-Ysard Puigbò and Paul Verschure
Information 2023, 14(8), 441; https://doi.org/10.3390/info14080441 - 4 Aug 2023
Viewed by 1402
Abstract
A major challenge in cognitive science and AI has been to understand how intelligent autonomous agents might acquire and predict the behavioral and mental states of other agents in the course of complex social interactions. How does such an agent model the goals, [...] Read more.
A major challenge in cognitive science and AI has been to understand how intelligent autonomous agents might acquire and predict the behavioral and mental states of other agents in the course of complex social interactions. How does such an agent model the goals, beliefs, and actions of other agents it interacts with? What are the computational principles to model a Theory of Mind (ToM)? Deep learning approaches to address these questions fall short of a better understanding of the problem. In part, this is due to the black-box nature of deep networks, wherein computational mechanisms of ToM are not readily revealed. Here, we consider alternative hypotheses seeking to model how the brain might realize a ToM. In particular, we propose embodied and situated agent models based on distributed adaptive control theory to predict the actions of other agents in five different game-theoretic tasks (Harmony Game, Hawk-Dove, Stag Hunt, Prisoner’s Dilemma, and Battle of the Exes). Our multi-layer control models implement top-down predictions from adaptive to reactive layers of control and bottom-up error feedback from reactive to adaptive layers. We test cooperative and competitive strategies among seven different agent models (cooperative, greedy, tit-for-tat, reinforcement-based, rational, predictive, and internal agents). We show that, compared to pure reinforcement-based strategies, probabilistic learning agents modeled on rational, predictive, and internal phenotypes perform better in game-theoretic metrics across tasks. The outlined autonomous multi-agent models might capture systems-level processes underlying a ToM and suggest architectural principles of ToM from a control-theoretic perspective. Full article
(This article belongs to the Special Issue Intelligent Agent and Multi-Agent System)
Show Figures

Figure 1

13 pages, 535 KiB  
Article
Reinforcement Learning-Based Hybrid Multi-Objective Optimization Algorithm Design
by Herbert Palm and Lorin Arndt
Information 2023, 14(5), 299; https://doi.org/10.3390/info14050299 - 22 May 2023
Cited by 3 | Viewed by 1756
Abstract
The multi-objective optimization (MOO) of complex systems remains a challenging task in engineering domains. The methodological approach of applying MOO algorithms to simulation-enabled models has established itself as a standard. Despite increasing in computational power, the effectiveness and efficiency of such algorithms, i.e., [...] Read more.
The multi-objective optimization (MOO) of complex systems remains a challenging task in engineering domains. The methodological approach of applying MOO algorithms to simulation-enabled models has established itself as a standard. Despite increasing in computational power, the effectiveness and efficiency of such algorithms, i.e., their ability to identify as many Pareto-optimal solutions as possible with as few simulation samples as possible, plays a decisive role. However, the question of which class of MOO algorithms is most effective or efficient with respect to which class of problems has not yet been resolved. To tackle this performance problem, hybrid optimization algorithms that combine multiple elementary search strategies have been proposed. Despite their potential, no systematic approach for selecting and combining elementary Pareto search strategies has yet been suggested. In this paper, we propose an approach for designing hybrid MOO algorithms that uses reinforcement learning (RL) techniques to train an intelligent agent for dynamically selecting and combining elementary MOO search strategies. We present both the fundamental RL-Based Hybrid MOO (RLhybMOO) methodology and an exemplary implementation applied to mathematical test functions. The results indicate a significant performance gain of intelligent agents over elementary and static hybrid search strategies, highlighting their ability to effectively and efficiently select algorithms. Full article
(This article belongs to the Special Issue Intelligent Agent and Multi-Agent System)
Show Figures

Figure 1

Review

Jump to: Research

47 pages, 864 KiB  
Review
Overview of Software Agent Platforms Available in 2023
by Zofia Wrona, Wojciech Buchwald, Maria Ganzha, Marcin Paprzycki, Florin Leon, Noman Noor and Constantin-Valentin Pal
Information 2023, 14(6), 348; https://doi.org/10.3390/info14060348 - 18 Jun 2023
Cited by 4 | Viewed by 5645
Abstract
Agent-based computing remains an active field of research with the goal of building (semi-)autonomous software for dynamic ecosystems. Today, this task should be realized using dedicated, specialized frameworks. Over almost 40 years, multiple agent platforms have been developed. While many of them have [...] Read more.
Agent-based computing remains an active field of research with the goal of building (semi-)autonomous software for dynamic ecosystems. Today, this task should be realized using dedicated, specialized frameworks. Over almost 40 years, multiple agent platforms have been developed. While many of them have been “abandoned”, others remain active, and new ones are constantly being released. This contribution presents a historical perspective on the domain and an up-to-date review of the existing agent platforms. It aims to serve as a reference point for anyone interested in developing agent systems. Therefore, the main characteristics of the included agent platforms are summarized, and selected links to projects where they have been used are provided. Furthermore, the described platforms are divided into general-purpose platforms and those targeting specific application domains. The focus of the contribution is on platforms that can be judged as being under active development. Information about “historical platforms” and platforms with an unclear status is included in a dedicated website accompanying this work. Full article
(This article belongs to the Special Issue Intelligent Agent and Multi-Agent System)
Show Figures

Figure 1

Back to TopTop