Plan Selection for Probabilistic BDI Agents

When an agent wants to fulfill its desires about the world, the agent usually has multiple plans to choose from and these plans have different pre-conditions and additional effects in addition to achieving its goals. Therefore, for further reasoning and interaction with the world, a plan selection strategy (usually based on plan cost estimation) is mandatory for an autonomous agent. This demand becomes even more critical when uncertainty on the observation of the world is taken into account, since in this case, we consider not only the costs of different plans, but also their chances of success estimated according to the agent's beliefs. In addition, when multiple goals are considered together, different plans achieving the goals can be conflicting on their preconditions (contexts) or the required resources. Hence a plan selection strategy should be able to choose a subset of plans that fulfills the maximum number of goals while maintaining context consistency and resource-tolerance among the chosen plans. To address the above two issues, in this paper we first propose several principles that a plan selection strategy should satisfy, and then we present selection strategies that stem from the principles, depending on whether a plan cost is taken into account. In addition, we also show that our selection strategy can partially recover intention revision.

Risk-aware Planning in BDI Agents

The ability of an autonomous agent to select rational actions is vital in enabling it to achieve its goals. To do so effectively in a high-stakes setting, the agent must be capable of considering the risk and potential reward of both immediate and future actions. In this paper we provide a novel method for calculating risk alongside utility in online planning algorithms. We integrate such a risk-aware planner with a BDI agent, allowing us to build agents that can set their risk aversion levels dynamically based on their changing beliefs about the environment. To guide the design of a risk-aware agent we propose a number of principles which such an agent should adhere to and show how our proposed framework satisfies these principles. Finally, we evaluate our approach and demonstrate that a dynamically risk-averse agent is capable of achieving a higher success rate than an agent that ignores risk, while obtaining a higher utility than an agent with a static risk attitude.

A graded BDI agent model to represent and reason about preferences

In this research note, we introduce a graded BDI agent development framework, g-BDI for short, that allows to build agents as multi-context systems that reason about three fundamental and graded mental attitudes (i.e. beliefs, desires and intentions). We propose a sound and complete logical framework for them and some logical extensions to accommodate slightly different views on desires. © 2011 Elsevier B.V. All rights reserved.

Incorporating PGMs into a BDI Architecture

In this paper, we present a hybrid BDI-PGM framework, in which PGMs (Probabilistic Graphical Models) are incorporated into a BDI (belief-desire-intention) architecture. This work is motivated by the need to address the scalability and noisy sensing issues in SCADA (Supervisory Control And Data Acquisition) systems. Our approach uses the incorporated PGMs to model the uncertainty reasoning and decision making processes of agents situated in a stochastic environment. In particular, we use Bayesian networks to reason about an agent’s beliefs about the environment based on its sensory observations, and select optimal plans according to the utilities of actions defined in influence diagrams. This approach takes the advantage of the scalability of the BDI architecture and the uncertainty reasoning capability of PGMs. We present a prototype of the proposed approach using a transit scenario to validate its effectiveness.

Can(Plan)+: Extending the Operational Semantics of the BDI Architecture to deal with Uncertain Information

The BDI architecture, where agents are modelled based on their beliefs, desires and intentions, provides a practical approach to develop large scale systems. However, it is not well suited to model complex Supervisory Control And Data Acquisition (SCADA) systems pervaded by uncertainty. In this paper we address this issue by extending the operational semantics of Can(Plan) into Can(Plan)+. We start by modelling the beliefs of an agent as a set of epistemic states where each state, possibly using a different representation, models part of the agent's beliefs. These epistemic states are stratified to make them commensurable and to reason about the uncertain beliefs of the agent. The syntax and semantics of a BDI agent are extended accordingly and we identify fragments with computationally efficient semantics. Finally, we examine how primitive actions are affected by uncertainty and we define an appropriate form of lookahead planning.

Context-dependent Combination of Sensor Information in Dempster-Shafer Theory for BDI

There has been much interest in the belief–desire–intention (BDI) agent-based model for developing scalable intelligent systems, e.g. using the AgentSpeak framework. However, reasoning from sensor information in these large-scale systems remains a significant challenge. For example, agents may be faced with information from heterogeneous sources which is uncertain and incomplete, while the sources themselves may be unreliable or conflicting. In order to derive meaningful conclusions, it is important that such information be correctly modelled and combined. In this paper, we choose to model uncertain sensor information in Dempster–Shafer (DS) theory. Unfortunately, as in other uncertainty theories, simple combination strategies in DS theory are often too restrictive (losing valuable information) or too permissive (resulting in ignorance). For this reason, we investigate how a context-dependent strategy originally defined for possibility theory can be adapted to DS theory. In particular, we use the notion of largely partially maximal consistent subsets (LPMCSes) to characterise the context for when to use Dempster’s original rule of combination and for when to resort to an alternative. To guide this process, we identify existing measures of similarity and conflict for finding LPMCSes along with quality of information heuristics to ensure that LPMCSes are formed around high-quality information. We then propose an intelligent sensor model for integrating this information into the AgentSpeak framework which is responsible for applying evidence propagation to construct compatible information, for performing context-dependent combination and for deriving beliefs for revising an agent’s belief base. Finally, we present a power grid scenario inspired by a real-world case study to demonstrate our work.

Hepatocyte growth factor (HGF) protects c-met-expressing Burkitt''s lymphoma cell lines from apoptotic death induced by DNA damaging agents

The relative sensitivity of neoplastic cells to DNA damaging agents is a key factor in cancer therapy. In this paper, we show that pretreatment of Burkitt's lymphoma cell lines expressing the c-met protooncogene with hepatocyte growth factor (HGF) protects them from death induced by DNA damaging agents commonly used in tumour therapy. This protection was observed in assays based on morphological assessment of apoptotic cells and DNA fragmentation assays. The protection was dose- and time-dependent — maximal protection requiring pre-incubation with 100 ng/ml HGF for 48 h. Western blotting analysis and flow cytometric studies revealed that HGF inhibited doxorubicin- and etoposide-induced decreases in the levels of the anti-apoptotic proteins Bcl-XL, and to a lesser extent Bcl-2, without inducing changes in the pro-apoptotic Bax protein. Overall, these studies suggest that the accumulation of HGF within the microenvironment of neoplastic cells may contribute to the development of a chemoresistant phenotype.