Accepted Papers

Below is the list of accepted papers:

Incremental Adaptative Organization for a Satellite Constellation, by Grégory Bonnet and Catherine Tessier

Physical agents, such as robots, are generally constrained in their communication capabilities. In a multi-agent system composed of physical agents, these constraints have a strong influence on the organization and the coordination mechanisms. Our multi-agent system is a satellite constellation, for which we propose a collaboration method based on an incremental coalition formation in order to optimize individual plans so as to satisfy collective objectives. This involves a communication protocol, a mutual knowledge notion and two coordination mechanisms: (1) an incentive to join coalitions and (2) coalition minimization. Results on a simulated satellite constellation are presented and discussed.

Decentralised Structural Adaptation in Agent Organisation, by Ramachandra Kota, Nicholas Gibbins and Nicholas R. Jennings

Autonomic computing is being advocated as a tool for maintaining and managing large and complex computing systems. Self-organising multi-agent systems provide a suitable paradigm for developing such autonomic systems. Towards this goal, we demonstrate a robust, decentralised approach for structural adaptation in explicitly modelled problem solving agent organisations. Our method is based on self-organisation principles and enables the agents to modify the organisational structure to achieve a better allocation of tasks across the organisation in a simulated task-solving environment. The agents forge and dissolve relations with other agents using their history of interactions as guidance. We empirically show that the efficiency of organisations using our approach is close to that of organisations having an omniscient central allocator and considerably better than static organisations or those changing the structure randomly.

Constraint Solving vs. Dynamic Reorganization for Task Allocation in Large Multi-agent Systems: Preliminary Results, by Ioannis Partsakoulakis, Kostas Stergiou and George Vouros

We consider the problem of efficiently allocating complex tasks consisting of interdependent subtasks in large multiagent systems. Complex tasks are modeled as distributed constraint optimization problems and agent teams that can potentially serve them are located using a search method that efficiently combines overlay networks of gateways and routing indices. In this paper we make a preliminary investigation of the trade-off between constraint reasoning and dynamic reorganization of agent teams aiming at maximizing the efficiency of the task allocation mechanism. We study experimentally a number of increasingly sophisticated constraint solving methods and compare them to a simple local search method that employs early reorganization. Preliminary results show that in heavily constrained cases the latter approach is more cost-effective.

Coordination in Adaptive Organisations: Extending Shared Plans with Knowledge Cultivation, by Kathleen Keogh, Liz Sonenberg and Wally Smith

Agent-based simulation can be used to investigate behavioural requirements, capabilities and strategies that might be helpful in complex, dynamic and adaptive situations, and can be used in training scenarios. In this paper, we study the requirements of coordination in complex unfolding scenarios in which agents may come and go and where there is no fixed organisational structure, with an eye to developing a simulation framework that can be part of a training system in the domain of emergency management. We describe an extension to the SharedPlans formalism required to support the sharing of knowledge about a dynamically unfolding situation, specifically: who is in the team, and who holds relevant knowledge. Our extension is based on a prior case study of a railway accident and a further analysis of the coordination and communication activities amongst the disaster management team during its recovery. We conclude that in addition to the obligations imposed by the standard SharedPlans framework, agents in complex unfolding scenarios also need knowledge cultivation processes to reason about the dynamic organisational structure and the changing world state.

Modeling Agent Adaptation in Games, by Joost Westra, Frank Dignum and Virginia Dignum

Almost all computer games that are currently created use fixed scenarios or simple fixed rules to define the course of the game. Using an approach like this will create very predictable and inflexible behavior of all the elements in the game. A lot of these games have the possibility to adjust the difficulty of the game, but only by defining a limited number of fixed scenarios. This has some disadvantages because not all people learn at similar rates. Different types of elements have different restrictions on the adjustments to guarantee a natural evolution of the game. Current research done on dynamic adjustability in games already makes it possible for different elements to adjust to the player. But these different elements are all adjusted by one centralized control system or by a multi-agent system using very centralized control. When the number of different elements increases this becomes an impractical approach. Our solution to this is to model the adjusting elements as autonomous units in a multi-agent system and use the agent organization system OperA to control the storyline of the game. We propose a fundamentally different way to specify the possible flow of the game while leaving enough flexibility for agents to adapt. For the implementation of the agents we suggest using adaptable BDI-agents by extending the 2APL programming language.

Autonomic Electronic Institutions' Self-Adaptation in Heterogeneous Agent Societies, by Eva Bou, Maite López-Sánchez, Juan Antonio Rodríguez-Aguilar and Jaime Simão Sichman

Electronic institutions (EIs) define the rules of the game in agent societies by fixing what agents are permitted and forbidden to do and under what circumstances. Autonomic Electronic Institutions (AEIs) adapt their rules to comply with their goals when regulating agent societies composed of varying populations of self-interested agents. We present a self-adaptation model based on Case-Based Reasoning (CBR) that allows an AEI to yield a dynamical answer to changing circumstances. In order to demonstrate adaptation empirically, we consider a traffic control scenario populated by heterogeneous agents. Within this setting, we demonstrate statistically that an AEI is able to adapt to different heterogeneous agent populations.

Modeling Feedback within MAS: A Systemic Approach to Organizational Dynamics, by Wolfgang Renz and Jan Sudeikat

Organization–oriented modeling approaches are established tools for Agent–Oriented Software Engineering (AOSE). Role and Group concepts are commonly used to design agent–based applications. These notions allow to partition and define static organizational struc tures and facilitate the description of agent behaviors in terms of role/group changing activities. Due to a growing interest in the construction of adaptive and self–organizing dynamics within MAS – i.e. applications that adjust their organizational structure at runtime – developers require tools for expressing the dynamics of MAS organizations, that result from individual agent activity and adaptiveness. In this paper we discuss how the macroscopic behavior of organizational structures can be modeled by relating systemic modeling techniques to MAS designs. Particularly the notions of causal links and causal loop diagrams are applied to express the timely behavior of role and group occupations. Corresponding modeling activities are facilitated by a graphical notation that highlights feedback loops. Since simulations are indispensable to examine complex, non–linear behaviors, we discuss how the systemic semantics can be translated into systems of stochastic process algebra terms, therefore enabling model simulation.

A Simulation Framework for the Investigation of Adaptive Behaviours in Largely Populated Building Evacuation Scenarios, by Daniele Gianni, Georgios Loukas and Erol Gelenbe

In an emergency scenario, civilians and emergency personnel have to continuously adapt their behaviour and make quick decisions to tackle unpredicted developments. Determining the optimal decisions and devising viable operational plans, while adapting to world changes, require systematic and accurate investigation of such systems. To effectively carry out such investigations in largely populated scenarios, we need a software framework that allows (i) reproducibility of the experiments, (ii) extendibility to diverse and unforeseen scenarios and (iii) distributed operation to allow the simulation of largely populated scenarios. We achieve all three requirements by developing an agent-based discrete-event simulation framework, and then building on top a Building Evacuation Simulator (BES), according to modern software engineering practices.

Cost of Control for Regulated Autonomy, by René Schumann, Andreas Lattner and Ingo Timm

Actual control and decision support systems rely more and more on distributed software systems. A central aspect in the design of these systems is the selection of an appropriate level of local autonomy of these subsystems in the decision making process. Most of the research focuses on the regulation of these entities towards an integration of centralized control strategies or a strict local autonomy, so far. Both aspects are margins of a scale and it can be expected that for certain applications a balance between autonomy and regulations has to be found. In our approach, we present first
steps towards a dynamic regulation of autonomous decision taking in order to design these adaptive decision support systems that try to solve the conflict between local autonomy and global system performance. Giving up a locally optimal strategy for the collective with global objectives leads to some costs. In this paper, we investigate the measurement of such costs and present experimental results how often central control is used instead of autonomous decision taking.

Modelling Actor Evolution in Agent-Based Simulations, by Aristama Roesli, Dominik Schmitz, Gerhard Lakemeyer and Matthias Jarke

Agent-based simulations have proven to be suitable to investigate many kinds of problems, especially in the field of social science. But to provide useful insights, the behaviour of the involved, simulated actors needs to reflect relevant features of the real world. In this paper, we address one particular aspect in this regard, namely the correct reflection of an actor's evolution during a simulation. Very often some knowledge exists about how an actor can evolve, for example, the typical development stages of entrepreneurs when investigating entrepreneurship networks. We propose to model this knowledge explicitly using evolution links between roles enriched with suitable conditions and extend i*, an agent- and goal-oriented modelling framework, thereby. We provide a mapping to the simulation environment ConGolog that serves as an intermediary approach between not providing change of behaviour at all and very open approaches to behaviour adaptation such as learning.

Adaptation of Voting Rules in Agent Societies, by Hugo Carr and Jeremy Pitt

We are concerned with multi-agent systems which are open, volatile, and decentralised, and which require collective use of a limited common resource. We define an institution to regulate access to the resource through a vote. We then investigate the adaptation of the voting rules to deal with the volatility of the agent presence, with the aim of achieving some semblance of `satisfaction' without depleting the resource. Progress towards implementation of an experimental testbed, for animating this type of multi-agent system, is described.