BDI Agent Research Articles

Experimental analysis of the effect of filtering perceptions in BDI agents

One of the BDI paradigm's major concerns is the lack of control over the agents' perceptions. Without having any form of goal-directed perceptions, agents may be flooded by irrelevant information thus causing an unjustified increase in processing time. This issue becomes critical when one needs to develop agents to be integrated with virtual environments or simulators, or even in the case of embedded agents, as robots. In order to provide greater control on the agents' perceptions and to reduce their time response, this work proposes to incorporate a filtering perception mechanism within the Jason interpreter, aiming to eliminate irrelevant perceptions in order to reduce processing time. To this end, some types of pre-defined filters proposed in the literature have been implemented and their effect experimentally evaluated in three different simulated and one embedded agent experiments. Using a full factorial experiment design, a technique used in computer systems performance analysis, it was shown that applying perception filters can reduce up to 80% of an agent's processing time, without significantly affecting its performance measured in terms of its utility function.

BDI2DoS: An application using collaborating BDI agents to combat DDoS attacks

Computer networks are critical to many tasks in our daily lives. Therefore, mechanisms that guarantee the resilience of the network must be provided. Different approaches have been proposed to address potential challenges to network operation, but they rely on rigid solutions, which work only in anticipated scenarios. To address this issue, this paper presents BDI2DoS, which is a multi-agent innovative application that ensures the resilience of networks against the widely known Distributed Denial-of-Service (DDoS) attack. We take an existing network resilience strategy based on event–condition–action (ECA) policies to combat DDoS attacks, and use it as a requirement to specify the behaviour that must emerge from the interaction among agents, which together are capable of detecting and remediating anomalies that are considered a DDoS threat, in a flexible way. Agents in our multi-agent system follow the widely used BDI architecture, and were implemented with the BDI4JADE agent platform. In order to evaluate the effectiveness of BDI2DoS, we used the PReSET resilience simulator and BDI4JADE to build an integrated testbed. Our experiments compare the effectiveness of the ECA and the BDI-based resilience strategies and show that the latter is more flexible and able to cope with problems that occur during the execution of the anticipated behaviour.

BDI agent testability revisited

Agent-based systems are deployed to solve a wide range of problems in a wide range of domains. Before software is deployed, it is important to obtain assurance that it will function correctly. Traditionally, this assurance is obtained by testing. However, there is an intuition that agents exhibit more complex behaviour than traditional software, which raises the question: how testable are agent systems? We focus on BDI agent programs, and analyse their testability with respect to the all edges test adequacy criterion (also known as “branch coverage”). Our results augment earlier results that considered the all paths criterion to provide a richer and more nuanced understanding of the testability of BDI agents. We show that the number of tests required with respect to the all edges criterion is much lower than that required with respect to the all paths criterion. We also show that, as for the previous analysis, BDI programs are harder to test than equivalently-sized procedural programs, even if exception handling is introduced. Overall, our conclusions lend strength to the earlier work, and motivate the need for work on formal methods for agent systems.

Toward Formal Modeling of Affective Agents in a BDI Architecture

Affective characteristics are crucial factors that influence human behavior, and often, the prevalence of either emotions or reason varies on each individual. We aim to facilitate the development of agents’ reasoning considering their affective characteristics. We first identify core processes in an affective BDI agent, and we integrate them into an affective agent architecture ( GenIA 3 ). These tasks include the extension of the BDI agent reasoning cycle to be compliant with the architecture, the extension of the agent language (Jason) to support affect-based reasoning, and the adjustment of the equilibrium between the agent’s affective and rational sides.

An Adjustable Autonomy Management Module for Multi-agent Systems

The design and development of software agents that perform adjustable autonomous actions in dynamic environments is a challenging research issue. Adjustable autonomous agents operate according to a spectrum of autonomy states. The autonomy states are dynamically adjusted to improve the agents’ performance and avoid autonomy surprises. For instance, the agents need to be guided or improvise when they encounter unplanned events. In this paper, we propose an Adjustable Autonomy Management (AAM) module for multi-agent systems. We integrate the AAM module within a BDI agent’s architecture to form an adjustable autonomous agent. The AAM module enables the agent to internally adjust its autonomy and handles external autonomy adjustments. We apply the agents in an autonomous Unmanned Aerial Vehicle (UAV) system. We test the UAV in a dynamic environment via experimenting aerial surveillance missions. The test results show that the agent-based UAV system successfully operates in the dynamic environment and performs the aerial surveillance missions. Subsequently, the AAM module is able to manage agents’ autonomy, reduce human workload and reduce adjustable autonomy disturbance.

BDI vs FSM Agents in Social Simulations for Raising Awareness in Disasters

Each summer in Australia, bushfires burn many hectares of forest, causing deaths, injuries, and destroying property. Agent-based simulation is a powerful tool to test various management strategies on a simulated population, and to raise awareness of the actual population behaviour. But valid results depend on realistic underlying models. This article describes two simulations of the Australian population's behaviour during bushfires designed in previous work, one based on a finite-state machine architecture, the other based on a belief-desire-intention agent architecture. It then proposes several contributions towards more realistic agent-based models of human behaviour: a methodology and tool for easily designing BDI models; a number of objective and subjective criteria for comparing agent-based models; a comparison of our two models along these criteria, showing that BDI provides better explanability and understandability of behaviour, makes models easier to extend, and is therefore best adapted; and a discussion of possible extensions of BDI models to further improve their realism.

A BDI Agent System for the Collaboration of the Unmanned Aerial Vehicle

A BDI Agent System for the Collaboration of the Unmanned Aerial Vehicle

Towards a Model of Open and Reliable Cognitive Multiagent Systems: Dealing with Trust and Emotions

Open multiagent systems are those in which the agents can enter or leave the system freely. In these systems any entity with unknown intention can occupy the environment. For this scenario trust and reputation mechanisms should be used to choose partners in order to request services or delegate tasks. Trust and reputation models have been proposed in the Multiagent Systems area as a way to assist agents to select good partners in order to improve interactions between them. Most of the trust and reputation models proposed in the literature take into account their functional aspects, but not how they affect the reasoning cycle of the agent. That is, under the perspective of the agent, a trust model is usually just a “black box” and the agents usually does not take into account their emotional state to make decisions as well as humans often do. As well as trust, agent’s emotions also have been studied with the aim of making the actions and reactions of the agents more like those of humans being in order to imitate their reasoning and decision making mechanisms. In this paper we analyse some proposed models found in the literature and propose a BDI and multi-context based agent model which includes emotional reasoning to lead trust and reputation in open multiagent systems.

Robust Execution of BDI Agent Programs by Exploiting Synergies Between Intentions

A key advantage the reactive planning approach adopted by BDI-based agents is the ability to recover from plan execution failures, and almost all BDI agent programming languages and platforms provide some form of failure handling mechanism. In general, these consist of simply choosing an alternative plan for the failed subgoal (e.g., JACK, Jadex). In this paper, we propose an alternative approach to recovering from execution failures that relies on exploiting positive interactions between an agent's intentions. A positive interaction occurs when the execution of an action in one intention assists the execution of actions in other intentions (e.g., by (re)establishing their preconditions). We have implemented our approach in a scheduling algorithm for BDI agents which we call SP. The results of a preliminary empirical evaluation of SP suggest our approach out-performs existing failure handling mechanisms used by state-of-the-art BDI languages. Moreover, the computational overhead of SP is modest.

Modelling and implementing modularised BDI agents with capability relationships

The BDI model is the foundation for one of the most widely used architectures to develop autonomous agents. Such model provides the concepts of beliefs, desires and intentions, which comprise the internal agent structure. Although much work has been done to support BDI agent development, there is lack of approaches that focus on modularisation of intra-agent software components. Given that agents often present a complex behaviour and, consequently, complex design and implementation, modularisation is a key to make the development of large-scale enterprise applications feasible. In this paper, we extend the concept of capability, which emerged to model BDI agent modules, by adding different types of relationships between them, namely association, composition and generalisation. Such relationships allow the development of BDI agent building blocks that can be combined so as to form agents, while hiding capability information as needed. Moreover, we present a modelling tool and implementation of the proposed relationships to not only provide the conceptual foundation of our approach but also enable its practical use. We show the effectiveness of our approach by refactoring an existing software product line implemented with BDI agents using our capability relationships.

Modelling and implementing modularised BDI agents with capability relationships

The BDI model is the foundation for one of the most widely used architectures to develop autonomous agents. Such model provides the concepts of beliefs, desires and intentions, which comprise the internal agent structure. Although much work has been done to support BDI agent development, there is lack of approaches that focus on modularisation of intra-agent software components. Given that agents often present a complex behaviour and, consequently, complex design and implementation, modularisation is a key to make the development of large-scale enterprise applications feasible. In this paper, we extend the concept of capability, which emerged to model BDI agent modules, by adding different types of relationships between them, namely association, composition and generalisation. Such relationships allow the development of BDI agent building blocks that can be combined so as to form agents, while hiding capability information as needed. Moreover, we present a modelling tool and implementation of the proposed relationships to not only provide the conceptual foundation of our approach but also enable its practical use. We show the effectiveness of our approach by refactoring an existing software product line implemented with BDI agents using our capability relationships.

Massive Cyber-Attacks Patterns Implemented with BDI Agents

This paper treats the new phenomenon of cyber-war as a new form of human violence between states. The concept of cyber-attack is a particular form of cybernetic war and it is one expensive in terms of cost. We tried to achieve a uniform approach of technological and sociological type using BDI agents and thus making the step to addressing computational sociology of cyberspace. We present two patterns of massive type attacks applicable in cyber space using BDI agents. The approach using BDI agents turned to be flexible and capable to model complex forms of cyber bullying that can occur between two countries in future.

Study & Analysis of Intention of a BDI Agent and its Performance

Study & Analysis of Intention of a BDI Agent and its Performance

Preference-based reasoning in BDI agent systems

An important feature of BDI agent systems is number of different ways in which an agent can achieve its goals. The choice of means to achieve the goal in made by the system at run time, depending on contextual information that is not available in advance. In this article, we explore ways that the user of an agent system can specify preferences which can be incorporated into the BDI execution process and used to guide the choices made. For example, a user of a travel system can specify a preferred airline, or a particular kind of accommodation, and the system will use this information to satisfy the goal and preferences, if possible. Preferences are specified in terms of properties of goals and resource usage, and are used to make two types of decisions: (a) select a plan when there is a choice and (b) determine the order in which subgoals of a plan should be pursued when their order is not fixed by design. We have implemented our preference framework in Jadex, and provide detailed case studies within the context of a holiday travel agent application.

Lateral-slip characteristics and combined model for vehicle tyres

It is important to model vehicular tyre movements and their lateral–slip characteristics because they affect safety while steering. In this study, we have used an artificial neural network (ANN) for modelling to solve issues such as insufficient learning adaptation and poor computability. However, because of the number of network nodes and layers and the superposition principle, the identification rate was low and node positions were uncertain. Hence, BDI–agent technology was introduced to optimise the training process of ANNs. We investigated and compared the performances of two models - the magic–formula–based and ANN–based models - for analysing tyre forces under combined representations of longitudinal and lateral forces. Our simulation results showed that the ANN–based model with BDI–agent performed better than the magic–formula–based model, and its mean error was low at 1.5%.

Requirement Analysis, Architectural Design and Formal Verification of a Multi-Agent Based University Information Management System

This paper presents an approach based on the analysis, design, and formal verification of a multi-agent based university Information Management System (IMS). University IMS accesses information, creates reports and facilitates teachers as well as students. An orchestrator agent manages the coordination between all agents. It also manages the database connectivity for the whole system. The proposed IMS is based on BDI agent architecture, which models the system based on belief, desire, and intentions. The correctness properties of safety and liveness are specified by First-order predicate logic.

A Quantum-BDI Model for Information Processing and Decision Making

This work aims to develop a novel BDI agent programming framework, which embeds the reasoning under uncertainty (probabilistic logic) and is capable of a realistic simulation of human reasoning. We claim that such a development can be addressed through the adoption of the mathematical and logical formalism derived from Quantum Mechanics: a scheme fulfilling the necessary requirements is described, useful for both the interpretation of some peculiarities in human behavior, and eventually the adoption of ‘quantum computing’ formalism for the agent programming. This last possibility could exploit the power of quantum parallelism in practical reasoning applications. Integration with the BDI paradigm enables the straightforward adoption of efficient learning algorithms and procedures, enhancing the behavior and adaptation of the agent to the environment.

A Dynamical Plan Revising for Ambient Systems

The proposed AgLOTOS formal specification language is dedicated to express BDI agent plans, according to the features and requirements of Ambient Intelligence (AmI). It offers a rich modular approach to express and compose elementary plans in order to execute them concurrently. We show how a plan is built automatically as a system of concurrent processes from the mental attitudes of the agent. In contrast to existing approaches, the plan is viewed as the realization of a whole set of partially ordered intentions. The AgLOTOS semantics accords with the possibility of updating some sub-plans on the fly, as the intention set of the BDI agent is revised.

Agent Decision-Making under Uncertainty: Towards a New E-BDI Agent Architecture Based on Immediate and Expected Emotions

 Abstract—Over the last decade, emotions have received considerable attention among scholars in agent oriented systems. In fact a large amount of computational models of emotions has been developed and a new generation of artificial agents has emerged to give rise to emotional agents, in particular the Emotional BDI (EBDI) agents. However, in spite of the several interesting studies that have been conducted to underline the role of emotions in decision-making, few works in the agent community have shed the light on the influences of both immediate and expected emotions to drive decision-making. In this context, we intend to propose a new conceptual model of EBDI agency that involves the interplay among immediate emotions, expected emotions and rational decisions of artificial agents.

A syntactic possibilistic belief change operator: Theory and empirical study

We propose a syntactic possibilistic belief-change operator, which operates on a belief base of necessity-valued formulas. Such a base may be regarded as a finite and compact encoding of a possibility distribution over a possibly infinite set of interpretations. The proposed operator is designed so that it behaves like a semantic possibilistic belief-change operator for BDI agents recently proposed in the literature. The equivalence of the semantic and syntactic operators is then proved. Experimental results are presented. The aim of these experiments is to demonstrate that the cost of belief revision expressed in terms of the number of entailment checks required as well as the size of the belief base do not explode as the number of new pieces of information formulas supplied increases.