Abstract. In this paper we describe a language for reasoning about actions that can be used for modelling and for programming rational agents. We propose a modal approach for reasoning about dynamic domains in a logic programming setting. Agent behavior is specified by means of complex actions which are defined using modal inclusion axioms. The language is able to handle knowledge producing actions as well as actions which remove information. The problem of reasoning about complex actions with incomplete knowledge is tackled and the temporal projection and planning problems is addressed; more specifically, a goal directed proof procedure is defined, which allows agents to reason about complex actions and to generate conditional plans. We give a non-monotonic solution for the frame problem by making use of persistency assumptions in the context of an abductive characterization. The language has been used for implementing an adaptive web-based system.

Agent theories and agent programs are two very different styles of specification of agent behavior. The former are declarative in nature, while the latter have an imperative flavor. In this paper, we combine ideas from both areas, yielding a powerful mode of agent specification that also gives the specifier a good deal of control over the complexity of the specified agent. In particular, we extend Shapiro et al.’s [16] agent theory to handle prioritized goals and then integrate it with the IndiGolog agent programming language. The result is a new IndiGolog construct that transforms a given nondeterministic, concurrent program δ into a new program δ ′ that can be described as a rational implementation of the original program, in the sense that δ ′ is an implementation of δ, and furthermore, δ ′ is the most rational of all implementations of δ relative to a given set of prioritized goals and the agent’s knowledge. With this construct, we can specify an agent that will attempt to achieve as many goals as possible in priority order even if the agent does not know of a plan that is guaranteed to achieve all the goals. In this case, the agent will select a plan that she thinks has the best chance of achieving the goals.

Much of the work in agent programming assumes an execution model where an agent has a knowledge base (KB) about the current state of the world, and makes decisions about what to do in terms of what is entailed or consistent with this KB. Deliberation or planning then would involve looking ahead and gauging what would be consistent or entailed at various stages by future KBs. We show that in the presence of sensing, this account of deliberation does not work properly, and propose an alternative that does.

The action language Golog has been applied successfully to the control of robots, among other things. Perhaps its greatest advantage is that a user can write programs which constrain the search for an executable plan in a flexible manner. However, when general planning is needed, Golog supports this only in principle, but does not measure up with state-of-the-art planners. In this paper we propose an integration of Golog and planning in the sense that planning problems, formulated as part of a Golog program, are solved by a modern planner during the execution of the program. Here we focus on the ADL subset of the plan language PDDL. First we show that the semantics of ADL can be understood as progression in the situation calculus, which underlies Golog, thus providing us with a correct embedding of ADL within Golog. We then show how Golog can be integrated with an existing ADL planner for closed-world initial databases and compare the performance of the resulting system with the original Golog. 1

In this paper we describe an agent-based infrastructure and toolkit to develop inter-operable, intelligent, multi-agent systems for Web service composition (WSC) and provisioning. Our toolkit is realized through an interface library (IG-JADE-PKSlib) that combines state of the art agentbased and planning technologies (i.e., the IndiGolog modelbased agent programming language, the JADE agent platform, and the PKS planning system). We show that each of these tools has its strengths and weaknesses, but combined together, they provide a very powerful toolkit. We argue that this infrastructure is particularly well suited for developing next generation Web services (WS) applications.

A recent trend in planning with incomplete information is to model the actions of a planning problem as nondeterministic transitions over the belief states of a planner, and to search for a plan that terminates in a desired goal state no matter how these transitions turn out. We show that this view of planning is fundamentally limited. Any plan that is successful by this criteria has an upper bound on the number of actions it can execute. Specifically, the account will not work when iterative plans are needed. We also show that by modifying the definition slightly, we obtain another account of planning that does work properly even for iterative plans. Although the argument is presented in an abstract form, we illustrate the issues using a simple concrete example.

Abstract. The Cognitive Agent Specification Language (CASL) is a framework for specifying and verifying complex communicating multiagent systems. In this paper, we extend CASL to incorporate a formal model of means-end reasoning suitable for a multiagent context. In particular, we define a simple model of cooperative ability, give a definition of rational plans, and show how an agent’s intentions play a role in determining her next actions. This bridges the gap between intentions to achieve a goal and intentions to act. We also define a notion of subjective plan execution and show that in the absence of interference, an agent that is able to achieve a goal, intends to do so, and is acting rationally and subjectively executing plans, will eventually achieve it. 1

In this paper, we develop a formal model of planning for an agent that is operating in a dynamic and incompletely known environment. We assume that both the agent’s task and the behavior of the agents in the environment are expressed as high-level nondeterministic concurrent programs in some agent programming language (APL). In this context, planning must produce a deterministic conditional plan for the agent that can be successfully executed against all possible executions of the environment program. We handle actions with nondeterministic effects, as well as sensing actions, by treating them as actions that trigger an environmental reaction that is not under the planning agent’s control. Our model of contingent planning is specified for a generic APL with a transition semantics. Within this model, we devise a general procedure for computing the contingent plans. We also show how the model can be instantiated in the situation calculus with programs for the agent and the environment expressed in ConGolog, and we describe an implementation of the planning mechanism in this case.

A thesis submitted in conformity with the requirements

In this paper we address the problem of planning by composing programs, rather than or in addition to primitive actions. The programs that form the building blocks of such plans can, themselves, contain both sensing and world-altering actions. Our work is primarily motivated by the problem of automated Web service composition, since Web services are programs that can sense and act. Our further motivation is to understand how to exploit macro-actions in existing operator-based planners that plan with sensing. We study this problem in the language of the situation calculus, appealing to Golog to represent our programs. To this end, we propose an offline execution semantics for Golog programs with sensing. We then propose a compilation method that transforms our action theory with programs into a new theory where programs are replaced by primitive actions. This enables us to use stateof-the-art, operator-based planning techniques to plan with programs that sense for a restricted but compelling class of programs. Finally, we discuss the applicability of these results to existing operator-based planners that support sensing and illustrate the computational advantage of planning with programs that sense via an experiment. The work presented here is cast in the situation calculus to facilitate formal analysis. Nevertheless, both the results and the algorithm can be trivially modified to take PDDL as input and output. This work has broad applicability to planning with programs or macro-actions with or without sensing. 1