Effects-Based Design of Robust Organizations

2013

Abstract:- Planning under uncertainty is a research subject that has attracted big attention in the past few years. A planner that operates under uncertainty must take into consideration that actions may have several different outcomes, some of which may be more desirable than others. The first developed decision support planners (DSP), use conventional Operations Research (OR) tools such as decision theory for developing Courses of Actions (COAs). Besides the conventional planning methods, other planning methods based on more sophisticated OR techniques such as the Markov Decision Processes (MDP) and Partially Observable Markov Decision Processes (POMDP) have been considered recently as promising approaches to use in developing decision support planners. The OR oriented planners, which can be considered for planning military operations, will be classified into three categories: i) conventional planners (based on replanning), ii) planners using the MDP paradigm, and iii) planners us...

In this paper we attempt to analyze the problem of centrally allocating tasks to selfinterested agents in environments characterized by high uncertainty. The analysis is based on equilibrium dynamics, taking into account the agents' short-term and long-term considerations. The model reflects a rich realism by allowing both tasks and agents to dynamically arrive in the environment. This is correlated with real life scenarios where agents take part in a continuous process while having only partial information concerning the different world states and other agents' capabilities. Based on the comprehensive analysis of the equilibrium we present innovative algorithms that significantly facilitate the extraction of the agents' strategies for any specific settings. The findings are illustrated computationally and compared with parallel mechanisms for allocating the tasks in a closed environment and in a centralized enforceable manner. Overall Overall System System Performance Performance Equilibrium Equilibrium Strategies Strategies Derive Derive MAS MAS Central Central Manager Manager Task Task Allocation Allocation Protocol Protocol Agent Agent' 's s Utility Utility Tasks Tasks Arrival Arrival Rates Rates Agents Agents Capabilities Capabilities

The Multi-Agent Distributed Goal Satisfaction (MADGS) system facilitates distributed mission planning and execution in complex dynamic environments with a focus on distributed goal planning and satisfaction and mixedinitiative interactions with the human user. By understanding the fundamental technical challenges faced by our commanders on and off the battlefield, we can help ease the burden of decision-making. MADGS lays the foundations for retrieving, analyzing, synthesizing, and disseminating information to commanders. In this paper, we present an overview of the MADGS architecture and discuss the key components that formed our initial prototype and testbed. Keywords Mobile multiagent systems. Mixed-initiative planning. Distributed mission planning and execution. Intelligent resource allocation. Agent oriented software engineering 1 Introduction and motivation Real-time air mission planning and execution occurs in a highly complex and dynamic environment where new E. Santos Jr.

2009

In this paper we present an approach to design an Organized Multiagent Systems (OMAS) for teamwork. We use a general formal model for OMAS that employs the notion of organizational mechanisms. The purpose of such mechanisms is influencing the behaviour of the agents towards more effectiveness with regard to some objectives. To achieve our goal we use Markov Decision Processes (MDPs) as a framework to design the organizational mechanisms.

We present a novel planning architecture that is composed of strategic and tactical layers. The complementary roles of these two layers are discussed, along with their relationship to one another, and a mathematical model is presented. The planning architecture is motivated by the requirements of dynamic systems, and it provides a clearly-defined optimality measure for an agent operating in a stochastic environment. Although the architecture can be applied in any domain, it is thus particularly well-suited to highly dynamic environments. Preliminary experimental results are presented.

Applied Intelligence, 2006

The Multi-Agent Distributed Goal Satisfaction (MADGS) system facilitates distributed mission planning and execution in complex dynamic environments with a focus on distributed goal planning and satisfaction and mixedinitiative interactions with the human user. By understanding the fundamental technical challenges faced by our commanders on and off the battlefield, we can help ease the burden of decision-making. MADGS lays the foundations for retrieving, analyzing, synthesizing, and disseminating information to commanders. In this paper, we present an overview of the MADGS architecture and discuss the key components that formed our initial prototype and testbed.

2005

Goal management is the process of recognizing or inferring goals of individual team members; abandoning goals that are no longer relevant; identifying and resolving conflicts among goals; and prioritizing goals consistently for optimal team collaboration and effective operations. A Markov decision process (MDP) approach is employed to maximize the probability of achieving the primary goals (a subset of all goals). We seek to address the computational adequacy of an MDP as a planning model by introducing novel problem domain-specific heuristic evaluation functions (HEF) to aid the search process. We employ the optimal AO* search and two suboptimal greedy search algorithms to solve the MDP problem. A comparison of these algorithms to the dynamic programming algorithm shows that computational complexity can be reduced substantially. In addition, we recognize that embedded in the MDP solution, there are a number of different action sequences by which a team's goals can be realized. That is, in achieving the aforementioned optimality criterion, we identify alternate sequences for accomplishing the primary goals.

Proceedings Eighth Conference on Artificial Intelligence for Applications

This paper presents a new approach to planning in dynamic and uncertain environments. Planning is viewed as a process in which an agent's long term goals are transformed into short term tasks and objectives, given the agent's strategy and the context of planning. The developed model allows for a dynamic balance between long term strategic planning and short term tactical planning.The notion of a strategy hierarchy is introduced to explicitly represent the process of strategy formulation and refinement. A combination of rules and plan scripts is used. Rules are used for representing domain knowledge and reasoning about strategic choices. Plan scripts are used for representing specific tasks and objectives. The uncertainty calculi of RUM/PRIMO [2], [6] are used for supporting reasoning under uncertainty. In the proposed model, it is also possible to achieve a seamless integration of case-based reasoning into the planning process. These ideas have been implemented in a system called MARS, which plans in the financial domain of mergers and acquisitions. I. PLANNING IN DYNAMIC AND UNCERTAIN DOMAINS N this section, we motivate the need for dynamic planning, I describe relevant prior research, and outline the structure of the paper.

2000

A thorough understanding of the relationships between a centre of gravity (COG) and its underlying critical capabilities and requirements is crucial to the development of a sound military plan. In this paper we describe the concepts underlying the COG Network Effects Tool (COGNET), which uses Bayesian (or causal probabilistic) networks to investigate this relationship structure. COGNET provides a visual representation of the causal structure and provides an effects-based analysis capability, which facilitates the determination of the critical vulnerabilities that have to be degraded or negated to influence the COG. Furthermore it provides a framework, which can serve as a knowledge base representing generic causal relationships to aid knowledge reusability and knowledge transfer.

2006

The naval domain is characterized by a dynamic environment. This requires constant adaptation of the organization, choosing between a wide variety of options. The consequences of the different options are difficult to foresee and hence, it is hard to judge which is best. This paper presents automated support for simulation, visualization, and validation of such adaptive multi-agent organisations. Generic simulation properties are specified using a formal modeling approach. Furthermore, results of a realistic case study are presented, and validated by means of properties obtained from naval experts. Finally, a tool is introduced that enables an insightful visualization of the simulation results.