The agentTool Project

Download agentTool
 

We are currently working on a new version of agentTool (aT3) to support the design of multiagent organizations ... check the aT3 page for details!!  We are expecting a beta release in late Fall 2005.  aT3 will be 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 FIPA compliant frameworks.

Overview

agentTool is a Java-based graphical development environment to help users analyze, design, and implement multiagent systems.  It is designed to support the Multiagent Systems Engineering (MaSE) methodology.  The system designer defines high-level system behavior graphically using the Multiagent Systems Engineering methodology.   The system design defines the types of agents in the system as well as the possible communications that may take place between agents.  This system-level specification is then refined for each type of agent in the system.  To refine an agent, the designer either selects or creates an agent architecture and then provides detailed behavioral specification for each component in the agent architecture. 

Once the system has been completely specified, the designer generates the code for the agent system.  It is at this point that the designer actually defines the underlying framework and any implementation specific communication and security protocols.   These are automatically built into the code during system synthesis.  To take advantage of pre-existing software components, agentTool also uses a library of AI components.  Each component has a formal definition which allows an agentTool designer to determine if the component meets the requirements of the specified agent.   In this way agentTool only has to synthesize code for components that are not already implementable via component reuse.

agentTool was developed using Java 1.3 and runs best on Windows platform, although it has been run successfully on other Java compatible platforms.  

Long Term Objectives

The long term objective of this research is to show that by using automated software synthesis techniques, intelligent agents can be developed that implement security and communication protocols in a provably correct manner. This research will also provide a mechanism to abstract the precise security and communication protocols so that the agent developer will not have to worry about them when specifying agent behavior. Also, assuming multiple security and communication protocols exist, an agent whose behavior is specified in such a system would be able to be generated using various combinations of security and communication requirements.

Approach

Our approach is to apply the general knowledge-based software synthesis approach discussed above to the intelligent agent domain. We have gained significant experience over the last few years attempting to apply knowledge-based synthesis in a domain-independent manner. Application of this knowledge to the intelligent agent domain will stress previously developed theories and necessitate the creation of new concepts and tools to solve domain-unique problems.

Technical Objectives

• Apply software synthesis techniques to construction of intelligent agents to ensure appropriate security and communications protocols are incorporated.
• Develop communication and security protocol models sufficient for use in automatic synthesis of agent-based software.
• Develop methods for hiding complex security and protocol issues from agent designers while ensuring their adherence.
• Implement a system to semi-automatically synthesize intelligent agents from high-level behavioral specifications that adhere to all required protocols.
• Integrate synthesized agents with other, possibly handcrafted, agents in an agent-based application.

Questions to be Explored:

• How to define security and communications models so they can be automatically incorporated into semi-automatically synthesized agents.
• How to specify intelligent agent behavior without requiring knowledge of detailed communication and security protocols.
• How to modify existing software synthesis processes to incorporate agent specific security and communication protocol issues.

Anticipated Results

We will show that intelligent agents can be semi-automatically synthesized from behavioral specifications that incorporate required communication and security protocols and that these agents can be successfully integrated into intelligent agent applications.

Sponsor

This research is sponsored by AFOSR.

Related Publications

• Scott A. DeLoach, Mark F. Wood and Clint H. Sparkman, Multiagent Systems Engineering, The International Journal of Software Engineering and Knowledge Engineering, Volume 11 no. 3, June 2001.
  
  
• Scott A. DeLoach.  Analysis and Design using MaSE and agentTool, Proceedings of the 12th Midwest Artificial Intelligence and Cognitive Science Conference (MAICS 2001).  Miami University, Oxford, Ohio,  March 31 - April 1, 2001.
  
  
• Marc J. Raphael & Scott A. DeLoach.  A Knowledge Base for Knowledge-Based Multiagent System Construction, National Aerospace and Electronics Conference (NAECON) to Dayton, OH, October 10-12, 2000.
  
  
• Scott A. DeLoach & Mark Wood, Developing Multiagent Systems with agentTool. The Seventh International Workshop on Agent Theories, Architectures, and Languages (ATAL-2000). Boston, MA, July 7-9, 2000.
  
  
• Lacey, Timothy H. & Scott A. DeLoach, Verification of Agent Behavioral Models.  The 2000 International Conference on Artificial Intelligence (IC-AI'2000).  June 26 - 29, 2000 Monte Carlo Resort, Las Vegas, Nevada
  
  
• Lacey, Timothy H. and Scott A. DeLoach, Automatic Verification of Multiagent Conversations,  the Eleventh Annual Midwest Artificial Intelligence and Cognitive Science Conference, April 15 - 16, 2000, University of Arkansas, Fayetteville