Search

Agent organizations are a good means to guarantee certain system objectives in the context of autonomous, self adapting agents. However, in highly dynamic environments such as in crisis management where different organizations have to cooperate flexibly and efficiently, the organizational structure itself should also be adaptable to the circumstances. In this paper we distinguish different types of context changes and the way organizations can adapt to such changes. We give a formal description –of both organizations and the changes– which facilitates proving that certain features remain stable throughout the life cycle of the organization or are justifiable changes. [ABSTRACT FROM AUTHOR]

Simulations for training can greatly benefit from BDI agents as virtual humans playing the role of key players. Learning to communicate effectively is a key aspect of training to command a team that is managing a crisis. In this paper, we present a goal-based dialogue system which has been applied to a navy fire-fighting incident simulation, training the commanding officer. The system enables more natural training of communication because agents utilize specialized communication goals. Communication is data-driven where the content is defined by a shared ontology of domain knowledge and tasks. The use of this ontology enables reuse of agent capabilities within the system and enables a tight coupling between the agents and the simulation. Categories and Subject Descriptors: I.2.0 [Artificial Intelligence]: General - Cognitive Simulation I.2.1 [Artificial Intelligence]: Applications and Expert Systems I.2.11 [Artificial Intelligence]: Distributed Artificial Intelligence - Intelligent Agents, Multiagent Systems I.6.3 [Simulation and Modeling]: Applications [ABSTRACT FROM AUTHOR]

The complexity of training situations requires teaching different skills to different trainees and in different situations. Current approaches of dynamic difficulty adjustment in games use a purely centralized approach for this adaptation. This becomes impractical if the complexity increases and especially if past actions of the non player characters need to be taken into account. Agents are increasingly used in serious game implementations as a means to reduce complexity and increase believability. Agents can be designed to adapt their behavior to different user requirements and situations. However, this leads to situations in which the lack of coordination between the agents makes it practically impossible to follow the intended storyline of the game and select suitable difficulties for the trainee. In this paper, we present a monitoring system for the coordination of the characters actions and adaptation to guarantee appropriate combinations of character actions that ensure the preservation of the storyline. In particular we propose an architecture for game design that introduces a monitoring module to check the development of user skills and direct coordinated agent adaptation. That is, agents propose possible courses of action that are fitting their role and context, and the monitor module uses this information together with its evaluation of user level and storyline progress to determine the most suitable combination of proposals. Categories and Subject Descriptors: I.2.11 [Artificial Intelligence]: Distributed Artificial Intelligence [ABSTRACT FROM AUTHOR]

Virtual characters in games operate in a social context involving other characters and possibly human players as well. If such socially situated virtual characters are to be considered believable to players, they should appear to adjust their behavior based on their (presumed) beliefs about the mental states of other characters. Autonomous BDI-based agents are suitable for modeling characters that base their actions on mental states attributed to other agents. In this paper, it is illustrated how agent-based characters can infer the mental state of other virtual characters by observing others΄ actions in the context of some scene in a role-playing game. Contextual information can be utilized in explanation and prediction of agents΄ behavior, and as such can form the basis for developing characters that appear to be socially aware. [ABSTRACT FROM AUTHOR]

Increasing complexity in serious games and the need to reuse and adapt games to different purposes and different user needs, requires distributed development approaches. The use of software agents has been advocated as a means to deal with the complexity of serious games. Current approaches to dynamic adjustability in games make it possible for different elements to adjust to the player. However, these approaches most use centralized control, which becomes impractical if the complexity and the number of adaptable elements increase. The serious games we are investigating are constructed using complex and independent subtasks that influence each other. In this paper, we propose a model for game adaptation that is guided by three main concerns: the trainee, the game objectives and the agents. In particular we focus on how the adaptation engine determines tasks to be adapted and how agents respond to such requests and modify their plans accordingly. [ABSTRACT FROM AUTHOR]

In virtual training scenarios, agent technology can be used to build a virtual tutor that assists a student during training. In a dialogue using argumentation schemes, the virtual tutor provides reasons to the students to explain why a particular action is the most sensible. The tutor determines the best action using practical reasoning. The justification of this action is selected based on the personality type of the student. This paper studies how agent technology could be used to make a virtual tutor that assists the student during the training. In particular, we study how the tutor can generate persuasive arguments for what the student should do. [ABSTRACT FROM AUTHOR]

The concept of cyclical influence between individuals and society is widely accepted, but hard to understand in all details. This paper proposes the use of three processes of social influence as a way to study the link between social and individual levels of abstraction. These processes are used to design an agent architecture which tries to provide explicit links to its social context. In order to detail the impact of the social influence, the architecture also includes personality and emotional aspects. [ABSTRACT FROM AUTHOR]

The process of introducing new public policies is a complex one in the sense that the behavior of society at the macro-level depends directly on the individual behavior of the people in that society and ongoing dynamics of the environment. It is at the micro-level that change is initiated, that policies effectively change the behavior of individuals. Since macro-models do not suffice, science has turned to develop and study agent-based simulations, i.e., micro-level models. In correspondence with the good scientific practice of parsimony, current ABSS models are based on agents with simple cognitive capabilities. However, the societies being modeled in policy making relate to real people with real needs and personalities, often of a multi-cultural composition. Those circumstances require the agents to be diversified to accommodate these facts. In this positioning paper, we propose an incrementally complex model for agent reasoning that can describe the influence of policies or comparable external influences on the behavior of agents. Starting from the BDI model for agent reasoning, we discuss the effect when personality and Maslow΄s hierarchy of needs are added to the model. Finally, we extend the model with a component that captures the cultural background and normative constitution of the agent. In the paper we show how these extensions affect the filtering of the desires and intentions of the agent and the willingness of the agent to modify its behavior in face of a new policy. This way, simulations can be made that support the differentiation of behaviors in multi-cultural societies, and thus can be made to support policy makers in their decisions. [ABSTRACT FROM AUTHOR]

Agents in an organization need to coordinate their actions in order to reach the organizational goals. Organizational models specify the desired behaviour in terms of roles, relations, norms and interactions. In this research, we show how organizational rules can be adopted by autonomous agents. We have developed a method to translate norms into reasoning rules for agents. However, since the agents are autonomous they will have their own reasoning rules next to the organizational rules. We propose a modular reasoning model to make organizational rules explicit. This allows for meta-reasoning about these rules. We show that this stimulates bottom-up dynamics in the organization. [ABSTRACT FROM AUTHOR]

In this paper we present mental state abduction, a technique for inferring the mental states (beliefs, goals) of BDI-based agents from observations of their actions. Abduced mental states are considered to be explanations of observed behavior, which is assumed to be part of a (goal-directed) plan. The observer, who attempts to explain an agent΄s behavior, is assumed to know all of the agent΄s behavioral rules that can account for observed actions. Three explanatory strategies are introduced, based on different perceptory conditions. [ABSTRACT FROM AUTHOR]