J-J.Ch. Meyer

An Integrated Modal Approach to Rational Agents

In this paper we give an overview of work we have done to provide a framework in which many aspects of rational agency are integrated. The various attitudes of a rational agent, viz. the informational as well as the motivational ones, are modelled in the framework by means of a variety of modal operators that are interpreted by means of possible worlds, as usual in modal logic. A main point here is that we incorporate all these modal operators into one model, so that in principle the various modal operators can be mixed to describe an agent’s complex attitudes.

Programming Normative Artifacts with Declarative Obligations and Prohibitions

Abstract—Normative concepts (e.g., obligations and prohibitions) are seen as a suitable tool for regulating the observable behavior of software agents. An enforcement mechanism – a process that detects when a norm is active, detects violations of the norms and handles these violations – is crucial for the practical use of norms in the construction of multi-agent systems. We propose a programming language for programming normative artifacts in terms of “declarative” norms referring to a state of the world (as opposed to “procedural” norms only referring to actions), and introduce the operational semantics of the norm enforcement mechanism.

Seeing is believing - And so are hearing and jumping

The formalisation of rational agents is a topic of continuing interest in Artificial Intelligence. Research on this subject has held the limelight ever since the pioneering work of Moore [1980; 1984] in which knowledge and actions are considered. Over the years important contributions have been made on both informational aspects like knowledge and belief [Halpern and Moses, 1992; Meyer and van der Hoek, 1995], and motivational 1 aspects like commitments and obligations [Cohen, 1990]. Recent developments include the work on agent-oriented programming [Shoham, 1993; Thomas, 1993], the Belief-Desire-Intention architecture [Rao and Georgeff, 1991; Rao and Georgeff, 1991a; Rao and Georgeff, 1993], logics for the specification and verification of multi-agent systems [Wooldridge, 1994; Wooldridge and Fisher, 1992], logics for agents with bounded rationality [Huang, 1994; Huang et al., 1992], and cognitive robotics [Lesperance et al., 1994; Levesque, 1994].

On programming KARO agents

After having defined the KARO logic for specifying intelligent agents (an amalgam of various modal logics to reason about both informational and motivational attitudes of agents) in earlier work we now turn to the question how to realise agents specified in the KARO framework. To this end we look at agent programming languages that we have defined, and investigate how programs in these languages can be linked to the KARO logic.

Autonomy and Coordination: Controlling External Influences on Decision Making

In order to achieve optimal results, an agents way of decision making might need to change according to the circumstances. One of the aspects an agent can adapt is the way it processes external events. Therewith it controls to what extend it is being influenced by external factors. We argue that influence control is a form of metareasoning. In this paper, we propose a model for metareasoning that allows the agent to select and prioritize heuristics for event processing. The model includes a performance measure based on goal achievement, a selection mechanism to select desired attitude and a control mechanism to adapt its attitude. We describe how to specify these components and we discuss the agents performance in a scenario involving coordination.

A coordination language for agents interacting in distributed plan-execute cycles

An architecture is proposed for agents to coordinate formation and execution of plans. Along with the architecture, a new coordination language is introduced that agents may employ to form temporary alliances for planning and plan execution. This language is based on ideas from constraint programming and a formal operational semantics is given for it.

Translations of Vocabularies in Systems of Communicating Agents

Abstract Abstract We develop a programming framework for systems of agents that interact with each other in a shared environment. We model the common environment as a mathematical structure and assign to each agent a part of this environment as its expertise. This expertise is given by a syntactical signature composed of a collection of relation and function symbols, which gives rise to a window on the environment. The agents observe and manipulate their individual windows and additionally exchange information about them along interconnecting communication channels. During such communication steps the agents may introduce new mental concepts to act as translations of concepts from foreign signatures that are referred to in the information received. Finally, the agents maintain a private belief base, which together with their window on the environment constitutes their mental state. This all results in an abstract programming language of communicating agents, which incorporates and generalizes traditional programming concepts from the field of concurrency theory. The syntax for the language is given, followed by a semantical description in terms of a transition system. This system enables the formal derivation of computations and gives rise to an operational semantics of the multi-agent programming language.