agentTool III

agentTool III (or aT3) is a project of the Multiagent & Cooperative Robotics (MACR) Laboratory at Kansas State University. agentTool is a Java-based graphical development environment to help users analyze, design, and implement multiagent systems.  It is designed to support the new Organization-based Multiagent Systems Engineering (O-MaSE) methodology.  The system designer defines high-level system behavior graphically using the flexible O-MaSE methodology.

The initial version of aT3 was released to the public in Fall 2007 as version 1.0.5.  aT3 is an Eclipse plug-in that will give the agent system designer unprecedented flexibility yet still retain the verification capabilities previously provided.  Eventually, aT3 will provide predictive performance metrics to allow the designer to make intelligent tradeoffs and will generate code for various multiagent platforms.

agentTool III is being developed using Java 1.5 and works with Eclipse 3.4 (Ganymede) through at least 3.7 (Indigo). The current release is available from our downloads page.

O-MaSE Models Provided

The current set of O-MaSE models that agentTool III supports includes:

• Goal Model. This is an AND/OR goal tree structure. Goals in the Goal Tree may have any number of attributes and goals may have precedence and triggering relationships between them and include the notion of events that occur during a goal and trigger other goals. [screenshot]

• Value-based Goal Model. Formally, the VGM is a tree whose nodes are value‐based goals (G) rooted at goal g0, which typically represents either the overall operational (business) goal of the network or the overall goal of system security. However, instead of achieving goals, the objective of a value‐based system is to maintain the current set of goals or the value of the system.

• Agent Model.  An Agent Class Model defines the agent classes and sub-organizations that will populate the organization. Agent Models are static models that may include agents, actors, organizations, roles, capabilities, protocols, and services. Relationships between the various entities include inheritance, possesses, plays, requires, and provides. [screenshot]

• Role Model.  A role model is a diagram representing all the roles in the organization along with the goals they are achieving and the interaction protocol existing in the organization. Role Models are static models that may include goals, roles, actors, capabilities, protocols, and services. Relationships between the various entities include inheritance, require, require, and provides. [screenshot]

• Organization Model. The Organizational Model is a diagram to show the interaction between the organization and the external actors. Organization Models are static models that may include actors, organizations, goals, protocols, and services. Relationships between the various entities include requires and provides. [screenshot]

• Protocol Model. A Protocol Model consists of Protocol Diagrams as specified in AUML. It describes sequences of messages sent between, roles, organizations, and external actors. It also includes alternative and looping structures. [screenshot]

• Plan Model. An agent plan model is based on Finite State Automata. Includes states and transitions. Use Capabilities/actions within states and on transitions. [screenshot]

• Capability-Action Model. PA Capability includes three levels of capabilities: Capabilities, Actions, and Operations. Capabilities may either be (exclusively) composed of a Plan or of other Capabilities/Actions. [screenshot]

• Domain Model. A domain model contains the definition of the environment in which the multiagent system is situated.  It consists of a set of Environment Objects (which may include agents), their attributes, and their relationships. The model may also include Environment Policies, which define the processes and principles that govern the multiagent system. [screenshot]

• Policy Model.  A policy model is a document containing all the policies applicable to the system. These will have some predefined formal notation. [screenshot]

Additional Features

aT³ also supports broader software engineering concepts:

• Verification Framework. Provides consistency checking within and between models. [screenshot]

• Code Generation. Code will be generated for various platforms. Currently, Initially, aT³ will support our internal Cooperative Robotic Organizational Simulator; however, future releases will include additional code generators.

• Metrics Computation. aT³ will provide a suite of metrics that helps developers choose, compute, and make changes to their system based on a design-time measurements. Many of the metrics will require state-based exploration and thus will include integration of the Bogor model checker .

• Process Engineering Support. The agentTool Process Editor (APE) supports the definition and verification of valid O-MaSE processes. The tool will eventually help configure aT³ to adhere to that process.

Sponsor

This research is sponsored by the National Science Foundation and the Air Force Office of Scientific Research.