Lin Padgham

Prometheus: a methodology for developing intelligent agents

As agents gain acceptance as a technology there is a growing need for practical methods for developing agent applications. This paper presents the Prometheus methodology, which has been developed over several years in collaboration with Agent Oriented Software. The methodology has been taught at industry workshops and university courses. It has proven effective in assisting developers to design, document, and build agent systems. Prometheus differs from existing methodologies in that it is a detailed and complete (start to end) methodology for developing intelligent agents which has evolved out of industrial and pedagogical experience. This paper describes the process and the products of the methodology illustrated by a running example.

Adding debugging support to the Prometheus methodology

This paper describes a debugger which uses the design artifacts of the Prometheus agent-oriented software engineering methodology to alert the developer testing the system, that a specification has been violated. Detailed information is provided regarding the error which can help the developer in locating its source. Interaction protocols specified during design, are converted to executable Petri net representations. The system can then be monitored at run time to identify situations which do not conform to specified protocols. A process for monitoring aspects of plan selection is also described. The paper then describes the Prometheus Design Tool, developed to support the Prometheus methodology, and presents a vision of an integrated development environment providing full life cycle support for the development of agent systems. The initial part of the paper provides a detailed summary of the Prometheus methodology and the artifacts on which the debugger is based.

Simplifying the Development of Intelligent Agents

Intelligent agents is a powerful Artificial Intelligence technology which shows considerable promise as a new paradigm for mainstream software development. However, despite their promise, intelligent agents are still scarce in the market place. A key reason for this is that developing intelligent agent software requires significant training and skill: a typical developer or undergraduate struggles to develop good agent systems using the Belief Desire Intention (BDI) model (or similar models). This paper identifies the concept set which we have found to be important in developing intelligent agent systems and the relationships between these concepts. This concept set was developed with the intention of being clearer, simpler, and easier to use than current approaches. We also describe briefly a (very simplified) example from one of the projects we have worked on (RoboRescue), illustrating the way in which these concepts are important in designing and developing intelligent software agents.

Tool support for agent development using the Prometheus methodology

We believe that tool support is very important for any methodology. In this paper we describe PDT (Prometheus design tool) which supports the design of an intelligent agent system using the Prometheus methodology. We describe how PDT supports the various stages of Prometheus through various means such as consistency checking, support for entity propagation, and hierarchical views. We also describe works that are currently in progress which involves the development of a plug-in for Eclipse with the aim of creating a single integrated development environment which would support the complete development cycle of an agent system from design to deployment.

A BDI agent programming language with failure handling, declarative goals, and planning

Agents are an important technology that have the potential to take over contemporary methods for analysing, designing, and implementing complex software. The Belief-Desire-Intention (BDI) agent paradigm has proven to be one of the major approaches to intelligent agent systems, both in academia and in industry. Typical BDI agent-oriented programming languages rely on user-provided “plan libraries” to achieve goals, and online context sensitive subgoal selection and expansion. These allow for the development of systems that are extremely flexible and responsive to the environment, and as a result, well suited for complex applications with (soft) real-time reasoning and control requirements. Nonetheless, complex decision making that goes beyond, but is compatible with, run-time context-dependent plan selection is one of the most natural and important next steps within this technology. In this paper we develop a typical BDI-style agent-oriented programming language that enhances usual BDI programming style with three distinguished features: declarative goals, look-ahead planning, and failure handling. First, an account that mixes both procedural and declarative aspects of goals is necessary in order to reason about important properties of goals and to decouple plans from what these plans are meant to achieve. Second, lookahead deliberation about the effects of one choice of expansion over another is clearly desirable or even mandatory in many circumstances so as to guarantee goal achievability and to avoid undesired situations. Finally, a failure handling mechanism, suitably integrated with both declarative goals and planning, is required in order to model an adequate level of commitment to goals, as well as to be consistent with most real BDI implemented systems.

Detecting & exploiting positive goal interaction in intelligent agents

Rational agents typically pursue multiple goals in parallel. However most existing agent systems do not have any infrastructure support for reasoning about either positive or negative interactions between goals. Negative interactions include such things as competition for resources, which if unrecognised can lead to unnecessary failure of both goals requiring the resource. Positive interactions include situations where there is potentially a common subgoal of two goals. This paper looks at mechanisms for identifying potential common subgoals, and attempting to schedule the actions of the agent to take advantage of this. Potential common subgoals are identified by maintaining summaries of definite and potential effects of goals and plans to achieve those goals. Template summaries for goal types are produced at compile time, while instance summaries are maintained and updated at execution time to allow the agent to choose and schedule its plans to take advantage of potential commonality where possible. This increases the ability of the agent to act in a rational manner, where rational is loosely defined as the sensible behaviour exhibited by humans.

Reframing social sustainability reporting : towards an engaged approach

Existing approaches to sustainability assessment are typically characterized as being either “top–down” or “bottom–up.” While top–down approaches are commonly adopted by businesses, bottom–up approaches are more often adopted by civil society organizations and communities. Top–down approaches clearly favor standardization and commensurability between other sustainability assessment efforts, to the potential exclusion of issues that really matter on the ground. Conversely, bottom–up approaches enable sustainability initiatives to speak directly to the concerns and issues of communities, but lack a basis for comparability. While there are clearly contexts in which one approach can be favored over another, it is equally desirable to develop mechanisms that mediate between both. In this paper, we outline a methodology for framing sustainability assessment and developing indicator sets that aim to bridge these two approaches. The methodology incorporates common components of bottom–up assessment: constituency-based engagement processes and opportunity to identify critical issues and indicators. At the same time, it uses the idea of a “knowledge base,” to help with the selection of standardized, top–down indicators. We briefly describe two projects where the aspects of the methodology have been trialed with urban governments and communities, and then present the methodology in full, with an accompanying description of a supporting software system.

Representation and reasoning for goals in BDI agents

A number of agent-oriented programming systems are based on a framework of beliefs, desires and intentions (BDI) and more explicitly on the BDI logic of Rao and Georgeff. In this logic, goals are a consistent set of desires, and this property is fundamental to the semantics of the logic. However, implementations based on this framework typically have no explicit representation of either desires or goals, and consequently no mechanisms for checking consistency. In this paper we address this gap between theory and practice by giving an explicit representation for a simple class of desires. The simplicity of this class makes it both straightforward and efficient to check for consistency. We provide a general framework for conflict resolution based on a preference ordering of sets of goals, and we illustrate how different rules for specifying consistent goal sets (corresponding to different preference orderings) relate to existing commitment strategies. We also report on some implementation experiments which confirm that the cost of consistency maintenance is not significant.

A Framework for Part-of Hierarchies in Terminological Logics

There is a growing recognition that part-whole hierarchies are a very general form of representation, widely used by humans in commonsense reasoning. This paper develops a terminological logic, and related inference mechanisms for representing and reasoning about composite concepts and individuals. A basic terminological logic language is extended with constructs for describing composite concepts in terms of their parts and the relationships between them. A part-of hierarchy is defined, based on the relationship of compositional inclusion. This part-of hierarchy is analogous to, but different from, the “is-a” hierarchy. Compositional inferencing is defined as a process which infers the existence of a whole, based on the existence of the required parts, where the parts are in the necessary relationship to each other. Three stable states are defined with respect to compositional inferencing - compositional extensions, credulous compositional extensions and skeptical compositional conclusions. This framework significantly enhances and is complementary to, knowledge representation and reasoning based on is-a hierarchies.

Model-Based Test Oracle Generation for Automated Unit Testing of Agent Systems

Software testing remains the most widely used approach to verification in industry today, consuming between 30-50 percent of the entire development cost. Test input selection for intelligent agents presents a problem due to the very fact that the agents are intended to operate robustly under conditions which developers did not consider and would therefore be unlikely to test. Using methods to automatically generate and execute tests is one way to provide coverage of many conditions without significantly increasing cost. However, one problem using automatic generation and execution of tests is the oracle problem: How can we automatically decide if observed program behavior is correct with respect to its specification? In this paper, we present a model-based oracle generation method for unit testing belief-desire-intention agents. We develop a fault model based on the features of the core units to capture the types of faults that may be encountered and define how to automatically generate a partial, passive oracle from the agent design models. We evaluate both the fault model and the oracle generation by testing 14 agent systems. Over 400 issues were raised, and these were analyzed to ascertain whether they represented genuine faults or were false positives. We found that over 70 percent of issues raised were indicative of problems in either the design or the code. Of the 19 checks performed by our oracle, faults were found by all but 5 of these checks. We also found that 8 out the 11 fault types identified in our fault model exhibited at least one fault. The evaluation indicates that the fault model is a productive conceptualization of the problems to be expected in agent unit testing and that the oracle is able to find a substantial number of such faults with relatively small overhead in terms of false positives.