Key to reinforcement learning in multi-agent systems is the ability to exploit the fact that agents only directly influence only a small subset of the other agents. Such loose couplings are often modelled using a graphical model: a coordination graph. Finding an (approximately) optimal joint action for a given coordination graph is therefore a central subroutine in cooperative multi-agent reinforcement learning (MARL). Much research in MARL focuses on how to gradually update the parameters of the coordination graph, whilst leaving the solving of the coordination graph up to a known typically exact and generic subroutine. However, exact methods { e.g., Variable Elimination { do not scale well, and generic methods do not exploit the MARL setting of gradually updating a coordination graph and recomputing the joint action to select. In this paper, we examine what happens if we use a heuristic method, i.e., local search, to select joint actions in MARL, and whether we can use outcome of this local search from a previous time-step to speed up and improve local search. We show empirically that by using local search, we can scale up to many agents and complex coordination graphs, and that by reusing joint actions from the previous time-step to initialise local search, we can both improve the quality of the joint actions found and the speed with which these joint actions are found.
LINK
Athor supplied : "This paper describes an agent-based architecture for domotics. This architecture is based on requirements about expandability and hardware independence. The heart of the system is a multi-agent system. This system is distributed over several platforms to open the possibility to tie the agents directly to the actuators, sensors and devices involved. This way a level of abstraction is created and all intelligence of the system as a whole is related to the agents involved. A proof of concept has been built and functions as expected. By implementing real and simulated devices and an easy to use graphical interface, all kind of compositions can be studied using this platform."
DOCUMENT
This chapter presents the currently not established and identifies design requirements for new systems to address this challenge and provide directions for possible improvement. As a result, this chapter introduces the concept of SamenMarkt®, a participatory system in which multi-agent system technology enables distributed price negotiation, distribution and communication between producers, retailers and consumers.
LINK
Author supplied: Abstract—The growing importance and impact of new technologies are changing many industries. This effect is especially noticeable in the manufacturing industry. This paper explores a practical implementation of a hybrid architecture for the newest generation of manufacturing systems. The papers starts with a proposition that envisions reconfigurable systems that work together autonomously to create Manufacturing as a Service (MaaS). It introduces a number of problems in this area and shows the requirements for an architecture that can be the main research platform to solve a number of these problems, including the need for safe and flexible system behaviour and the ability to reconfigure with limited interference to other systems within the manufacturing environment. The paper highlights the infrastructure and architecture itself that can support the requirements to solve the mentioned problems in the future. A concept system named Grid Manufacturing is then introduced that shows both the hardware and software systems to handle the challenges. The paper then moves towards the design of the architecture and introduces all systems involved, including the specific hardware platforms that will be controlled by the software platform called REXOS (Reconfigurable EQuipletS Operating System). The design choices are provided that show why it has become a hybrid platform that uses Java Agent Development Framework (JADE) and Robot Operating System (ROS). Finally, to validate REXOS, the performance is measured and discussed, which shows that REXOS can be used as a practical basis for more specific research for robust autonomous reconfigurable systems and application in industry 4.0. This paper shows practical examples of how to successfully combine several technologies that are meant to lead to a faster adoption and a better business case for autonomous and reconfigurable systems in industry.
DOCUMENT
Efficiency of city logistics activities suffers due to conflicting personal preferences and distributed decision making by multiple city logistics stakeholders. This is exacerbated by interdependency of city logistics activities, decision making with limited information and stakeholders’ preference for personal objectives over system efficiency. Accordingly, the key to understanding the causes of inefficiency in the city logistics domain is understanding the interaction between heterogeneous stakeholders of the system. With the capabilities of representing a system in a natural and flexible way, agent based modelling (ABM) is a promising alternative for the city logistics domain. This research focuses on developing a framework for the successful implementation of the ABM approach for the city logistics domain. The framework includes various elements – a multi-perspective semantic data model (i.e. ontology) and its validation, the development of an agent base model using this ontology, and a validation approach for the agent-based model. Conclusively, the framework shows that a rigorous course can be taken to successfully implement agent based modelling approach for the city logistics domain.
DOCUMENT
This paper describes an agent-based software infrastructure for agile industrial production. This production is done on special devices called equiplets. A grid of these equiplets connected by a fast network is capable of producing a variety of different products in parallel. The multi-agent-based underlying systems uses two kinds of agents: an agent representing the product and an agent representing the equiplet.
MULTIFILE
With artificial intelligence (AI) systems entering our working and leisure environments with increasing adaptation and learning capabilities, new opportunities arise for developing hybrid (human-AI) intelligence (HI) systems, comprising new ways of collaboration. However, there is not yet a structured way of specifying design solutions of collaboration for hybrid intelligence (HI) systems and there is a lack of best practices shared across application domains. We address this gap by investigating the generalization of specific design solutions into design patterns that can be shared and applied in different contexts. We present a human-centered bottom-up approach for the specification of design solutions and their abstraction into team design patterns. We apply the proposed approach for 4 concrete HI use cases and show the successful extraction of team design patterns that are generalizable, providing re-usable design components across various domains. This work advances previous research on team design patterns and designing applications of HI systems.
MULTIFILE
Sustainable and Agile manufacturing is expected of future generation manufacturing systems. The goal is to create scalable, reconfigurable and adaptable manufacturing systems which are able to produce a range of products without new investments into new manufacturing equipment. This requires a new approach with a combination of high performance software and intelligent systems. Other case studies have used hybrid and intelligent systems in software before. However, they were mainly used to improve the logistic processes and are not commonly used within the hardware control loop. This paper introduces a case study on flexible and hybrid software architecture, which uses prototype manufacturing machines called equiplets. These systems should be applicable for the industry and are able to dynamically adapt to changes in the product as well as changes in the manufacturing systems. This is done by creating self-configurable machines which use intelligent control software, based on agent technology and computer vision. The requirements and resulting technologies are discussed using simple reasoning and analysis, leading to a basic design of a software control system, which is based on a hybrid distributed control system
DOCUMENT
The methodology of biomimicry design thinking is based on and builds upon the overarching patterns that all life abides by. “Cultivating cooperative relationships” within an ecosystem is one such pattern we as humans can learn from to nurture our own mutualistic and symbiotic relationships. While form and process translations from biology to design have proven accessible by students learning biomimicry, the realm of translating biological functions in a systematic approach has proven to be more difficult. This study examines how higher education students can approach the gap that many companies in transition are struggling with today; that of thinking within the closed loops of their own ecosystem, to do good without damaging the system itself. Design students should be able to assess and advise on product design choices within such systems after graduation. We know when tackling a design challenge, teams have difficulties sifting through the mass of information they encounter, and many obstacles are encountered by students and their professional clients when trying to implement systems thinking into their design process. While biomimicry offers guidelines and methodology, there is insufficient research on complex, systems-level problem solving that systems thinking biomimicry requires. This study looks at factors found in course exercises, through student surveys and interviews that helped (novice) professionals initiate systems thinking methods as part of their strategy. The steps found in this research show characteristics from student responses and matching educational steps which enabled them to develop their own approach to challenges in a systems thinking manner. Experiences from the 2022 cohort of the semester “Design with Nature” within the Industrial Design Engineering program at The Hague University of Applied Sciences in the Netherlands have shown that the mixing and matching of connected biological design strategies to understand integrating functions and relationships within a human system is a promising first step. Stevens LL, Whitehead C, Singhal A. Cultivating Cooperative Relationships: Identifying Learning Gaps When Teaching Students Systems Thinking Biomimicry. Biomimetics. 2022; 7(4):184. https://doi.org/10.3390/biomimetics7040184
DOCUMENT
Author supplied: The production system described in this paper in an implementation of an agile agent-based production system. This system is designed to meet the requirements of modern production, where short time to market, requirementdriven production and low cost small quantity production are important issues. The production is done on special devices called equiplets. A grid of these equiplets connected by a fast network is capable of producing a variety of different products in parallel. The multi-agent-based software infrastructure is responsible for the agile manufacturing. A product agent is responsible for the production of a single product and equiplet agents will perform the production steps to assemble the product. This paper describes this multiagent-based production system with the focus on the product agent. Presented at EUMAS 2013 ( 11th European Workshop on Multi-Agent Systems) , At Toulouse.
DOCUMENT
