Download presentation
Presentation is loading. Please wait.
Published byΝύξ Ελευθερίου Modified over 6 years ago
1
MAGE: A Multi-Agent Environment for Humanized Systems
Zhongzhi Shi Fen Lin 30 October 2005
2
Outline Introduction Agent Model Overview of MAGE
Agent Supporting Environment Agent Development Environment Examples Conclusions 11/28/2018 Zhongzhi Shi MAGE
3
Introduction Humanized systems are man-made systems Closer To Humans,
Exhibit behaviors characteristic of natural living system 11/28/2018 Zhongzhi Shi MAGE
4
Artificial Fish 11/28/2018 Zhongzhi Shi MAGE
5
HUMANOID 11/28/2018 Zhongzhi Shi MAGE
6
Robocup 11/28/2018 Zhongzhi Shi MAGE
7
Introduction Agent plays an important role in humanized systems
Agent exhibit human intelligence Multi-agent systems facilitate the realization of humanized systems adaptive,open,distributed systems based on negotiation or cooperation Multi-agent systems gains more and more interest in both the research area and the industry migrate from the research laboratories to the software industry 11/28/2018 Zhongzhi Shi MAGE
8
Outline Introduction Agent Model Overview of MAGE
Agent Supporting Environment Agent Development Environment Examples Conclusions 11/28/2018 Zhongzhi Shi MAGE
9
Agent Model Focuses on mental state description of agent.
Here we use dynamic description logic (DDL) for agent representation with a clear and formally defined semantics. By combining the features of knowledge representation and reasoning both in static and dynamic domains, DDL is effective and significant for agent reasoning and programming 11/28/2018 Zhongzhi Shi MAGE
10
Dynamic Description Logic
K B TBox(Scheme) Man = Human ⊓ Male Happy-father = Human ⊓ ∃ Has-child.Female⊓ … Abox(Data) John: Happy-father <John,Mary> : Has-child Reasoning Interface 11/28/2018 Zhongzhi Shi MAGE
11
Belief Belief knowledge base K = <T, S, B > where
T: set of axioms about concepts and their definitions S: set of causality constraint axioms that are used to keep the consistence and complement of belief knowledge base B : current state 11/28/2018 Zhongzhi Shi MAGE
12
Belief Revision AddBelief(F, B) { F′ Extend(F); Foreach ϕF′ do
If B Then B B { }; B′ Extend(B∪F); If Consistent(B′) Then Return B′; Else{ Let {ψ,ψ} = ConflictSet(B′); If ψB Then Return B′ {ψ}; Else If ψB Then Return B′ {ψ}; Else Return error; }} 11/28/2018 Zhongzhi Shi MAGE
13
Goal Goal Set of goal G (1) AG,A basic action;
(2) If L,then achieve( )G; (3) If L,then ?G; (4)若 1, 2G,则1;2G,1⋃2G,1*G; 11/28/2018 Zhongzhi Shi MAGE
14
Goal Target Set T: (1) If is a basic action α,then T =E P,where E is the result of α, P is the preconditions α; (2) If has the form achieve(),then T = { }; (3) If is a test action ,then T = 。 11/28/2018 Zhongzhi Shi MAGE
15
Goal Generation Rule 1, …, n where
1, …, n are assertion formulas, is a goal When the assertions 1, …, n hold, agent will generates a new goal . 11/28/2018 Zhongzhi Shi MAGE
16
Planning Static Planning, Planning rule: 1 | 2
Where 1G, 2G, L. 1: rule header 2: rule body, : rule guard 11/28/2018 Zhongzhi Shi MAGE
17
Dynamic planning Possible state sets of
Let be a basic goal, T be the target set of and α be an action. For the result E of, , if there exists T such that E, i.e. a sub-goal of can be realized by the execution of α, then is E called a possible state set of . If the target formulas T of appear in the result E1,…,En of actions α1,…,αn at the same time, then all E1,…,En of α1,…,αn are called possible state sets of . 11/28/2018 Zhongzhi Shi MAGE
18
Dynamic planning Planning algorithm Order of sub-goals
Assuming that is a goal and can be divided into two sub-goals a and b, then a has precedence over b if a must be achieved before b in order to achieve , which is denotes as a≺b. Otherwise there is no need to achieve b before b and this can be denoted by a⊀b. Planning algorithm 11/28/2018 Zhongzhi Shi MAGE
19
Planning Algorithm Plan(,B) //Plan for goal , where B is current beliefs Begin If is executable in B and can realize Then Return B; Else Search for all sub-goals of ; Compute the priorities of these sub-goals; Order sub-goals and record them as 1,…, n; For = 1 to n do B = Plan(,B); Enqueue(, P); B = (B P) E; Return B; EndIF End 11/28/2018 Zhongzhi Shi MAGE
20
Outline Introduction Agent Model Overview of MAGE
Agent Supporting Environment Agent Development Environment Examples Conclusions 11/28/2018 Zhongzhi Shi MAGE
21
Overview of MAGE MAGE (Multi-Agent Environment) is an integrated tool suite for constructing multi-agent systems MAGE is designed to be compliant with FIPA, its agents is comply with FIPA Agent Management Specification. MAGE provide friendly and easy-to-use human-computer interface through visual programming paradigm and pick-and-choose principle. 11/28/2018 Zhongzhi Shi MAGE
22
Overview of MAGE MAGE supports the entire process of AOSE
Analysis, Design , Development and Deployment MAGE consists of agent supporting environment and agent development environment. AUMP, VAStudio and Agent Supporting Environment 11/28/2018 Zhongzhi Shi MAGE
23
Agent Architecture Agent Kernel Agent Architecture Engine Plug-INs
Function Component Function interface sensor Engine Communicator Plug-in Manager Plug-INs Reasoning Scheduling Negotiation Resource Database Task Database Cooperation Agent Kernel Others Agent Architecture 11/28/2018 Zhongzhi Shi MAGE
24
Agent Architecture Agent Kernel. Sensor perceives the outside world.
Function Module Interface makes an effect to the outside. Communicator handles communications between the agent and other agents. Coordinator makes decisions concerning the agent’s goals, and it is also responsible for coordinating the agent interactions with other agents using given co-ordination protocols and strategies. Scheduler plans the agent’s tasks based on decisions taken by the Co-ordination Resource Database maintains a list of resources that are owned by and available to the agent. Task Database provides logical descriptions of tasks known to the agent. Plug-In Manager manages the components provided by MAGE or by users that can be plugged into agent kernel. 11/28/2018 Zhongzhi Shi MAGE
25
Work Flow of MAGE Work Flow of MAGE Requirement Analysis
System Development System Deployment System Design Behaviour Library Agent Society Agent Library Work Flow of MAGE 11/28/2018 Zhongzhi Shi MAGE
26
Work Flow of MAGE Analysis: developing an understanding of the system and its structure. Design: transform the analysis models into a sufficiently low level of abstraction that traditional design techniques Development: constructing a functional solution to the problem. Here we divide development phase into three steps: building behaviours, building agents and building system. Deployment: actualizing the solution to the real problem in the given domain and managing the runtime environment. 11/28/2018 Zhongzhi Shi MAGE
27
Agent development environment Design and Programming Tool VAStudio
MAGE Framework Agent development environment Modeling Tool AUMP Design and Programming Tool VAStudio Agent supporting environment MAGE Framework 11/28/2018 Zhongzhi Shi MAGE
28
MAGE Framework AUMP is designed for system analysis and design stages
VAStudio is for system design, development and deployment stages Agent Supporting Environment provides agent running environment 11/28/2018 Zhongzhi Shi MAGE
29
Outline Introduction Agent Model Overview of MAGE
Agent Supporting Environment Agent Development Environment Conclusions Examples 11/28/2018 Zhongzhi Shi MAGE
30
Agent Supporting Platform
Software MAGE Agent Library Directory Facilitator Agent Management System Agent 主体 主体 Function component Message Transport System (MTS) Other Agent Platforms Message Transport System (MTS) 11/28/2018 Zhongzhi Shi MAGE
31
Agent Supporting Platform
AMS offers white pages services to other agents. DF provides yellow pages services to other agents. MTS is the default communication method between agents on different agent platforms. Agent is the fundamental actor in MAGE which combines one or more service capabilities into a unified and integrated execution model that may include access to external software, human users and communications facilities. Software describes all non-agent, executable collections of instructions accessible through an agent. Moreover, two auxiliary modules are provided to support designing agent systems: Agent Library and Function Component. 11/28/2018 Zhongzhi Shi MAGE
32
Outline Introduction Agent Model Overview of MAGE
Agent Supporting Environment Agent Development Environment Examples Conclusions 11/28/2018 Zhongzhi Shi MAGE
33
MAS hierarchical Model
11/28/2018 Zhongzhi Shi MAGE
34
MAS hierarchical Model
A component can be almost any reusable unit A behaviour is the capability unit of agent An agent can be seen as a software entity An agent society is composed of agents, including the interaction between them, the protocol they use, the ontology they adopt. two ways to develop a multi-agent system: top-down approach, bottom-up approach. 11/28/2018 Zhongzhi Shi MAGE
35
VAStudio Architecture
11/28/2018 Zhongzhi Shi MAGE
36
VAStudio Architecture
VAStudio design module supports behaviour and agent design Flow Chart mode, FSM mode, clone mode and ADL mode VAStudio programming module supports code editing, compiling and debugging. MAS running environment interface supports the important API using in VAStudio by MAGE agent supporting environment Toolkits: behaviour editor, protocol editor, ontology editor and strategy editor. behaviour library and agent library makes it possible to reuse resources. 11/28/2018 Zhongzhi Shi MAGE
37
Emotion Agent Planning Sensor Belief Emotion Inf. Emotion KB
Environment Emotion KB Rational Inf. Effector Intention Desire 11/28/2018 Zhongzhi Shi MAGE
38
MAGE Comparison 11/28/2018 Zhongzhi Shi MAGE Phase Agent Platform
AgentBuilder Jack Zeus MAGE Analysis Completeness: ★★★ ★★★★ ★★★★★ Applicability: Complexity: Reusability Design ★ ★★ Development ★★★★★★ Deployment ★★★ ★★★ ★★★ ★★★ 11/28/2018 Zhongzhi Shi MAGE
39
Outline Introduction Agent Model Overview of MAGE
Agent Supporting Environment Agent Development Environment Application Examples Conclusions 11/28/2018 Zhongzhi Shi MAGE
40
Applications: spider URL Database URL agent DF Spider agent
Web page Index agent DF 11/28/2018 Zhongzhi Shi MAGE
41
Examples 11/28/2018 Zhongzhi Shi MAGE
42
Data Mining Workflow of Execution Engine
A typical data mining workflow for classification Normalization Phase AttributeReduct Phase Discretization Phase Training Set and Testing Set Training Set and Testing Set Models and Predicting Results Testing Indexes step1 step1 step1 step1 stepN1 stepN2 stepN3 stepN4 Preprocessing Evaluating Training & Testing
43
Execution Process of Data Mining Workflow
Coordination Process of Agents MAGE Container-1 Schedule Agent Monitor Agent Mining Task Agent 1 Mining Task Agent 2 Mining Task Agent φ ………… MAGE Container-2 Schedule Agent Monitor Agent Mining Task Agent 1 Mining Task Agent 2 Mining Task Agent ψ ………… 11/28/2018 Zhongzhi Shi MAGE
44
Grid-Based Emergency Interactive System GEIS
计算机网络系统 城市应急联动 和社会综合服务系统 有无线通信系统 联动中心终端设备 RDBMS ODBMS IFR… IVR CTI 呼叫管理中间件 支撑平台 RMS… 业务汇聚 MIS MAP CAD 中心辅助系统 其它… 防汛 医疗 市政 公安 专业应用系统 数据库系统 调度 大屏 监控 GPS 外部系统 指挥决策系统 预案管理 情报管理 指挥控制… 11/28/2018 Zhongzhi Shi MAGE
45
Conclusions MAGE, a multi-agent environment for humanized systems, with a suit of tools to support agent-oriented requirement analysis, design, development and deployment DDL for Agent Model MAGE provides learning behavior library with rich learning algorithms for autonomous mental development of humanized systems Future work is going to development of mind agent with emotion and evolution which will improve the performance of humanized systems dramatically. 11/28/2018 Zhongzhi Shi MAGE
46
Welcome Intelligence Science http://www.intsci.ac.cn/ 11/28/2018
Intelligence Science 11/28/2018 Zhongzhi Shi MAGE
47
Thank You Question! 11/28/2018 Zhongzhi Shi MAGE
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.