Kurt Schelfthout

Infrastructures for the environment of multiagent systems

The notion of environment is receiving an increasing attention in the development of multiagent applications. This is witnessed by the emergence of a number of infrastructures providing agent designers with useful means to develop the agent environment, and thus to structure an effective multiagent application. In this paper we analyse the role and features of such infrastructures, and survey some relevant examples. We endorse a general viewpoint where the environment of a multiagent system is seen as a set of basic bricks we call environment abstractions, which (i) provide agents with services useful for achieving individual and social goals, and (ii) are supported by some underlying software infrastructure managing their creation and exploitation. Accordingly, we focus the survey on the opportunities that environment infrastructures provide to system designers when developing multiagent applications.

Applying the ATAM to an architecture for decentralized control of a transportation system

For two years, we have been involved in a challenging project to develop a new architecture for an industrial transportation system. The motivating quality attributes to develop this innovative architecture were flexibility and openness. Taking these quality attributes into account, we proposed a decentralized architecture using multiagent systems (MASs). A MAS consists of multiple autonomous entities that coordinate with each other to achieve decentralized control. The typical advantages attributed to such decentralized architecture are flexibility and openness, the motivating quality attributes to apply MAS in this case. The Architecture Tradeoff Analysis Method (ATAM) was used to provide insights wether our architecture meets the expected flexibility and openness, and to identify tradeoffs with other quality attributes. Applying the ATAM proved to be a valuable experience. One of the main outcome of applying the ATAM was the identification of a tradeoff between flexibility and communication load that results from the use of a decentralized architecture. This paper describes our experiences in applying the ATAM to a MAS architecture, containing both the main outcomes of the evaluation and a critical reflection on the ATAM itself.

Exploiting a virtual environment in a real-world application

In situated multi-agent systems (situated MASs), agents are explicitly placed in an environment. A situated agent does not not use long-term planning to decide what action sequence should be executed, but selects actions on the basis of its current position, the world it perceives and limited internal state. Situated agents exploit the environment to coordinate their behavior and to reach a common goal. In a recent project, we applied situated MASs to the control of an automated transportation system that uses automatic guided vehicles (AGVs) to transport loads in a warehouse. In contrast to traditional approaches where the AGVs are controlled by a central server, in this project we model the AGVs as agents in a situated MAS, aiming to improve flexibility and openness. Since the physical environment of AGVs is very restricted, it offers little opportunities for agents to use the environment. We introduce a virtual environment for agents to live in. This virtual environment (1) offers a medium that agents can use to exchange information and coordinate their behavior, and (2) serves as a suitable abstraction to shield low-level physical processing from the AGV agents. Since the only infrastructure available to the AGVs is a wireless network, the virtual environment is necessarily distributed over the AGVs. Synchronization of the state of the virtual environment is provided by ObjectPlaces, a middleware infrastructure that offers support to exchange and share information among nodes in mobile and ad-hoc networks. In this paper, we demonstrate how the environment is used creatively in the design of a MAS solution, helping to manage the complexity of engineering a complex real-world application.

Decentralized control of automatic guided vehicles: applying multi-agent systems in practice

An automatic guided vehicle (AGV) transportation system is a fully automated system that provides logistic services in an industrial environment such as a warehouse or a factory. Traditionally, the AGVs that execute the transportation tasks are controlled by a central server via wireless communication. In a joint effort between Egemin, an industrial manufacturer of AGV transportation systems, and DistriNet Labs research at the Katholieke Universiteit Leuven, we developed an innovative decentralized architecture for controlling AGVs. The driving motivations behind decentralizing the control of AGVs were new and future quality requirements such as flexibility and openness. At the software architectural level, the AGV control system is structured as a multi-agent system; the detailed design and implementation is object-oriented. In this paper, we report our experiences with developing the agent-based control system for AGVs. Starting from system requirements, we give an overview of the software architecture and we zoom in on a number of concrete functionalities. We reflect on our experiences and report lessons learned from applying multi-agent systems for real-world AGV control.

Middleware for protocol-based coordination in dynamic networks

Pervasive and ad hoc computing applications are frequently deployed in dynamic networks. Due to mobility of the computing nodes, their unreliability, or a limited communication range, at any time a node may enter or leave an interaction between a group of application components. Middleware approaches have been proposed to deal with these dynamics, by supporting the dissemination (or gathering) of information in dynamic networks. In our experience however, applications frequently need to execute a complete protocol to coordinate. Existing middleware can then be used as a discovery mechanism, but offers no support for handling the protocol itself. This paper presents a middleware model that enables an easier implementation of distributed protocols that need to take into account the continuously changing context in the dynamic network. It uses roles as a first order abstraction, handles the distributed instantiation of roles in an interaction session, and maintains the session as nodes in the mobile network move. We describe our experience with applying the middleware in a case study on a system of automatic guided vehicles.

ObjectPlaces: An Environment for Situated Multi-Agent Systems

Situated multi-agent systems are MAS with an explicit representation of the environment, which agents can observe and in which agents can act. An appropriate middleware platform should support the construction and manipulation of such an environment. We require three properties from the environment: (1) observableness: an agent must be able to get a current view on its context; (2) activity: the environment updates the agentýs current view; (3) dynamism: the environment can change without direct in- fluence from the agents. ObjectPlaces is a tuplespace-like middleware that offers an event-based interface and a concept of dynamic objects to support these requirements.

A pheromone-based coordination mechanism applied in peer-to-peer

In this paper, we discuss the principle of synthetic pheromones, which we view as a high level coordination mechanism suitable for scalable, distributed systems, such as peer to peer systems. We present a software abstraction for the application of synthetic pheromones, building on an existing coordination mechanism, objectspaces. The coordination principle is evaluated on the problem of search in a file-sharing P2P system.

Architectural design of a distributed application with autonomic quality requirements

An autonomic system is essentially characterized by quality requirements that specify that the system should be able to adapt itself (configure optimize, heal, etc.) under varying circumstances and situations. These quality requirements call for an architecture centric software engineering approach. In this paper, we discuss and illustrate the architectural design of a complex real-world distributed application with autonomic quality requirements. In particular, we present an architecture with autonomous entities (agents) for managing warehouse logistics. We illustrate how the subsequent architectural decisions are guided by a reference architecture for situated multi-agent systems on the one hand, and by functional and quality requirements of the application on the other hand.

Towards adaptive role selection for behavior-based agents

This paper presents a model for adaptive agents. The model describes the behavior of an agent as a graph of roles, in short a behavior graph. Links between roles provide conditions that determine whether the agent can switch roles. The behavior graph is assigned at design time, however adaptive role selection takes place at runtime. Adaptivity is achieved through factors in the links of the behavior graph. A factor models a property of the agent or its perceived environment. When an agent can switch roles via different links, the factors determine the role the agent will switch to. By analyzing the effects of its performed actions the agent is able to adjust the value of specific factors, adapting the selection of roles in line with the changing circumstances. Models for adaptive agents typically describe how an agent dynamically selects a behavior (or action) based on the calculation of a probability value as a function of the observed state for each individual behavior (or action). In contrast, the model we propose aims to dynamically adapt logical relations between different behaviors (called roles here) in order to dynamically form paths of behaviors (i.e. sequences of roles) that are suitable in the current state. To verify the model we applied it to the Packet-World. In the paper we discuss simulation results that show how the model enables the agents in the Packet-World to adapt their behavior to changes in the environment.

Coordination middleware for decentralized applications in dynamic networks

The Ph.D. work presented in this paper describes novel middleware abstractions for the support of decentralized applications in dynamic networks. Decentralized applications are characterized by the absence of an application component that has global control; a network is dynamic if its composition changes frequently and unexpectedly over time. In such a domain, application components are necessarily spread over the network nodes and need to coordinate among each other to achieve the applications functionality. The goal of the Ph.D. research is to support this coordination by suitable middleware abstractions. We describe two prototypes that were built with this goal in mind. First, a middleware supporting views, abstractions for representing and maintaining context information in a mobile ad hoc network is presented. A second middleware, that enables protocol-based interaction in mobile networks by supporting roles as a first order abstraction, is described. The application of this second middleware in a real world case study of automatic guided vehicle control is presented, showing its usefulness. Ongoing and further research in this area is discussed.