Demos
We present a simulator to compare different approaches toorganisation-centred MAS adaptation in a peer-to-peer (P2P)scenario. In particular, we describe our approach to MASadaptation (2-LAMA), the P2P sharing network case studyand the software we built to evaluate different alternatives. A Simulator for Organisation-Centred MAS Adaptation in P2P Sharing Networks
Dora the Explorer is a mobile robot with a sense of curiosity and a drive to explore its world. Given an incompletetour of an indoor environment, Dora is driven by internalmotivations to probe the gaps in her spatial knowledge. Sheactively explores regions of space which she hasn’t previouslyvisited but which she expects will lead her to further unexplored space. She will also attempt to determine the categories of rooms through active visual search for functionallyimportant objects, and through ontology-driven inference onthe results of this search. Dora The Explorer: A Motivated Robot
There is a chronic lack of shared application domains totest the research models and agent architectures on areaslike negotiation, argumentation, trust and reputation. Inthis demostration we introduce such a friendly testbed calleddipGame that can be used for all such purposes. The testbedis based on the Diplomacy Game due to its lack of randommoves and because of the essential role that negotiation andthe relationships between the players play in the game. Thetestbed may also profit from the existence of a community ofbot (player software agent) developers and a large numberof human players that would provide data for experiments.We offer the infrastructure, including a bot, and make itfreely available to the MAS community. dipGame: A Testbed for Multiagent Systems
A multi-agent coordination and control system is deployed forcontrolling multiple interacting agricultural vehicles involved inthe crop harvesting process. Crops are gathered by combineharvesters. The harvested product is transferred to one or moretractors every time the combine harvester’s storage capacity isreached (figure 1). Good cooperation between the combineharvesters and the tractors is important for successfullycompleting the harvesting process. The multi-agent system allowsconcurrent planning and execution of the process, aiming toincrease efficiency of the vehicles and improve cooperationbetween them. The planning is performed by short-termoperational forecasting. The system provides detailed instructionsand guidance to the operators of the individual vehicles (combineharvesters and tractors) by means of a graphical user interface.The state updates from the agricultural vehicles are considered bythe multi-agent system, which dynamically updates the controlinstructions. In this way the control instructions for the agriculturevehicles remain valid and effective throughout the process. Multi-agent Coordination and Control System for Multi-Vehicle Agricultural Operations
EMFGormas is a new approach for modeling Service-orientedOpen Multiagent Systems using the MDA Eclipse Technology. It offers a CASE tool based on an unified meta-modelfor engineering large-scale open systems in which the constituent entities interact among them by means of services. EMFGormas: A CASE Tool for Developing Service-Oriented Open MAS
Masev (MAs Software engineering EValuation framework)is an online application that allows analyzing and comparingmethods, techniques and environments for developing MAS.Moreover, Masev allows the evaluation of how these methods and tools support the development of OrganizationalMAS and Service-oriented MAS. Finally, Masev provides aquantitative comparison based on a simple metric whose parameters can be modified by the users. An Evaluation Tool for Multiagent Development Techniques
In this paper we present a system that automatically generates a 3D virtual world from a virtual institution (VI)specification, namely of activities that will take place in thevirtual world. A virtual institution is an interaction environment where both human and software agents participatein a joint 3D virtual world. A VI is visualized as a 3D building composed of rooms representing the different activitiesdefined in its specification. To support the generation ofthe 3D scene, we propose an extension of shape grammarscalled virtual world grammar (VWG). A VWG incorporatesinformation about the definition of the activities, propertiesof the different shapes (i.e spaces), functions to evaluate andvalidate the generated design, and a set of heuristics to guidethe generation process. A definition and execution of Virtual World Grammars and a 3D transformation mechanismare integrated in a so called Virtual World Builder Toolkit (VWBT). A Virtual World Builder Toolkit
We demonstrate a “Companion” ECA, which is able to provideadvice and support to the user, taking into account emotionsexpressed by her through dialogue. The integration of allrequired multimodal I/O components is based on interactionstrategies defining the shape of dialogue, on the ECA’s responsetimes, and on the underlying affective strategy. The systemsupports free conversation on an everyday life scenario in whichthe user comments her day at the office. How Was Your Day? A Companion ECA
This paper describes the THOMAS framework, a useful framework for the development of virtual organizations, on thebasis of a service-based approach. THOMAS: A Service-Oriented Framework For Virtual Organizations
High performance computing is essential for simulating complex problems using agent-based modelling (ABM). Researcheare hindered by complexities of porting models on parallelplatforms and time taken to run large simulations on a singlemachine. This paper presents FLAME framework, the onlysupercomputing framework which automatically producesparallelisable code to execute on different parallel hardwarearchitectures. FLAME’s inherent parallelism allows largenumber of agents to be simulated in less time than comparable simulation frameworks. The framework also handlesthe parallelisation of model code allowing modellers to runsimulations on number of supported architectures.The framework has been well tested in various disciplineslike biology and economics projects yielding successful research results like project EURACE, where the Europeaneconomy was modelled using agents. More recently FLAMEhas been ported to consumer NVIDIA Graphics ProcessingUnits (GPUs) allowing parallel performance equal to thatof grid architectures with the ability to perform real timevisualisation. FLAME: Simulating Large Populations of Agents on Parallel Hardware Architectures
Service-oriented systems are becoming more and more nodes in adigital, dynamic ecosystem requiring the identification and establishment of flexible, spontaneous collaboration activities betweenservices. This requires new engineering approaches that can integrate new functionalities and behaviours into existing running systems of active, distributed and interdependent processes. Here, wepresent the approach of the FP7 ALIVE1 project to the use of formalmodels of coordination and organisation mechanisms to deliver aflexible, high-level means to describe the structure of interactionsbetween services in the environment. We propose to build on thecurrent activities in service-oriented engineering by defining threelevels: (i) An organisational level models the organisational structure of executing and interlinked services and the context aroundthem. (ii) A coordination level provides flexible ways to model interaction between the services. (iii) These two levels connect withexisting (semantic) Web services, which contain semantic descriptions to make components aware of their social context and of therules of engagement with other services. ALIVE: An Agent-Based Framework for Dynamic and Robust Service-Oriented Applications
This demo illustrates MAS-DisCoSim 4 PDP, a testbed environment for evaluating distributed multi-agent system solutions to pickup and delivery problems (PDPs). PDPs arewell-studied problems in operational research, and providea quite suitable context for testing and evaluating a widevariety of distributed solutions, using particular reasoningand/or coordination mechanisms. The testbed provides areusable simulation framework, and a particular model forPDP environments.The demo illustrates the testbed by showing two MAS-based solutions, one based on gradient field coordination, asecond solution based on the coordination mechanism called‘delegate MAS’. The demo is interactive: a user can selectany part of a map, downloaded from the OpenStreetMapwebsite, can run various experiments, and can change - during simulation - various problem and solution characteristicsand observe their effct. MAS-DisCoSim 4 PDP: A Testbed for Multi-Agent Solutions to PDPs
We present an agent-based system for modeling, analyzingand reasoning in the maritime domain with the emphasis ondetecting, anticipating and preventing illegal activities, suchas contemporary maritime piracy. At the core of the systemis a data-driven agent-based simulation which combines arange of sources of crime-related real-world data with simulated operation of thousands of vessels of different typesin order to create a rich model of maritime activity. Thesimulation is integrated with a number of advanced reasoning methods for analyzing illegal activities and for planningactive counter-measures. In combination with experimentsupport tools and a powerful user frontend based on GoogleEarth, the testbed provides a complete environment for thedevelopment and evaluation of anti-maritime-crime methodsbased on the multi-agent approach. AgentC: Agent-based Testbed for Adversarial Modeling and Reasoning in the Maritime Domain
Bushfire BLOCKS is a modular, distributed, agent-basedsimulation for exploring and informing bushfire responsestrategies. Separate independent modules capture the firespread, evacuation of traffic and human behaviour such asthe decision whether to remain at property. These moduleshave been developed largely independently, using paradigmsand data models appropriate to their individual purpose.Modules are integrated into a single global simulationvia central services for managing time advancement andaccess to shared variables. Execution is distributed, withmodules running on different machines, whilst a custominterface enables viewing and control from one screen. Theunderlying architecture of this system facilitates extensionof individual modules and addition of new modules, withlimited alteration to other modules. Bushfire BLOCKS: A Modular Agent-Based Simulation
We demonstrate an approach for collision- and oscillation-free navigation of multiple robots or virtual agents amongsteach other. Each entity acts independently and uses onlyboth the position and velocity of nearby entities to predicttheir future trajectories in order to avoid collisions. Entitiestake into account that the other entities are responding tothem likewise to prevent oscillations. Independent Navigation of Multiple Robots and Virtual Agents
The RoboCup Rescue Simulation league has been a partof RoboCup since 2001. Starting in March 2009 a majoroverhaul of the platform’s features has been conducted. Wepresent a summary of the platform, major improvements inthe last 12 months, and a demonstration of the new platformhighlighting it’s new interactive features. The RoboCup Rescue Simulation Platform
The Intelligent Decentralised Energy-Aware Systems (iDEaS) projectat the University of Southampton (see www.ideasproject.info) isdeveloping and demonstrating the application of intelligent agentswithin the smart grid; a future vision of an electricity distributionnetwork capable of autonomous and intelligent configuration, robust and flexible operation, and two-way information flow betweenconsumers and suppliers. In this demonstration, we show howagent technologies can assist in realising three key components ofthis vision, specifically: (i) how a home energy management agentis capable of monitoring, visualising and coordinating energy usewithin the home, (ii) how micro-storage of electricity, coordinatedby intelligent agents, can flattern demand across the grid and reduceboth costs and carbon emissions, and (iii) how trading agents operating within a novel market mechanism can effectively and robustlydistribute energy within the smart grid whilst explicitly accountingfor the capacity constraints of the transmission lines. Intelligent Agents for the Smart Grid
A major characteristic of combat is the need to quickly adapt to new situations. Providing support and protection for a convoy is challenging since it requires coordination between the diferent participants to eciently allocate the protecting resources. In a combat situation, such as when the convoy is under attack, forces have no spare time to spend. Every second is extremely valuable, and context switching between different tasks is not desirable. Moreover, decision making in a stressful situation, such as combat, tends to be far from ecient. Another aspect that is central to distributed combat is the requirement to keep communication usage as low as possible so that the adversary cannot apply electronic warfare technologies to intercept messages and introduce interference that would disallow distributed collaboration. Another reason is that by reducing the amount of unmanned air vehicle (UAV) communication we provide more reliable communication bandwidth for urgent needs.
We propose innovative use of teams of small and inexpensive UAVs to autonomously allocate convoy protection tasks and robustly adapt their behavior to the changing environment. We envision a dynamically self-organized protective shield based on large-scale multiagent concepts. Each UAV will be controlled by an onboard agent that will sense, reason, and act autonomously according to the changing environment. Communication will be done by observation of the behavior of other UAVs and convoy vehicles rather than by message exchange. To support scalability, many large-scale multiagent paradigms suggest interacting with only a limited number of agents. Those agents should be carefully selected to provide an ecient and valuable solution. In this work we use concepts inspired by the Reynolds algorithm and physics-based theories. Each agent in our scalable solution is capable of reasoning about the behavior of some UAV agents and some vehicles that are part of the protective convoy. A range of interaction is introduced to construct a closed area in which any other UAV or convoy vehicle is notable by the agent. To provide further exibility and more general applicability, we allow separate interaction ranges for agent-UAV and agent-convoy vehicles.Convoy Protection by Self-Organized Teams of UAVs
The Restaurant Game demonstrates an end-to-end system that captures and generates social behavior for virtual agents. Over 15,000 people have played The Restaurant Game, and we have developed a system to automatically learn patterns of interaction and dialogue from logs of their gameplay sessions. These patterns guide a case-based planning system, which generates behavior and dialogue for a virtual customer or waitress who can interact with a human, or with another agent. The Restaurant Game demonstrates a first step toward empowering non-programmers to realize socially intelligent characters for a wide range of applications.Capturing and Generating Social Behavior with The Restaurant Game
As computational learning agents continue to improve their ability to learn sequential decision-making tasks, a central but largely unfulfilled goal is to deploy these agents in real-world domains in which they interact with humans and make decisions that affect our lives. People will want such interactive agents to be able to perform tasks for which the agent's original developers could not prepare it. Thus it will be imperative to develop agents that can learn from natural methods of communication. The teaching technique of shaping is one such method. In this context, we define shaping as training an agent through signals of positive and negative reinforcement. In a shaping scenario, a human trainer observes an agent and reinforces its behavior through push-buttons, spoken word (''yes'' or ''no''), facial expression, or any other signal that can be converted to a scalar signal of approval or disapproval. We treat shaping as a specific mode of knowledge transfer, distinct from (and probably complementary to) other natural methods of communication, including programming by demonstration and advice-giving. The key challenge before us is to create agents that can be shaped effectively.Training a Tetris Agent via Interactive Shaping: A Demonstration of the TAMER Framework
The Prometheus Design Tool (PDT) is a graphical tool that is used to design a Multi-Agent System following the Prometheus Methodology. This paper describes the latest version of PDT which is now integrated into the Eclipse platform, enabling the users to accomplish the full development life-cycle of an agent-oriented application in one IDE and also inherit the rich set of product development features that Eclipse provides. This version of PDT also aims to support simpler integration with tools from other AOSE methodologies where appropriate.Eclipse-based Prometheus Design Tool
Teams of people need to coordinate in real-time in many dynamic and uncertain domains. Examples include disaster rescue, hospital triage, and military operations. It is possible to develop plan a priori, but many parts of these plans must be left unspecified because people won't know exactly what needs to be done until they are executing the plan in the field. Additionally, requirements and tasks can evolve during execution. Our work addresses a fundamental multi-agent systems endeavor of creating decision support systems that help humans perform better in these domains. The technical challenges to compute good solutions for these problems have been well documented.Decision-Support for Real-Time Multi-Agent Coordination