This research shows rational decision-making and coordination among antiair units whose mission is to defend a specified territory from a number of attacking missiles. The defense units have to coordinate and decide which missiles they are to attempt to intercept, given the characteristics of the threat, and given what they can expect of the other defense units. Our approach is based upon maximizing the overall survival prospects of the attacked territory, which gives the highest priority to the attacking missiles that pose the greatest threat. We assume that threat evaluation can be achieved by considering such attributes as the altitude of the missile and the size of its warhead. Further, the defense units consider the hit probability with which their interceptors would be effective against each of the hostile missiles.

Each of the two defense units can launch three interceptors and are faced with an attack by six missiles named as A, B, C, D, E, and F, respectively. In these example runs, missile warhead sizes for A..F are 470, 410, 350, 370, 420, 450. Each case below is executed over 100 times, and the quality of the coordination acheived is measured as the combined expected tonnage of the missiles that penetrated the defense and damaged the protected territory (since each interceptor has a kill probability of less than one, the residual probability that the missile has not been destroyed is multiplied by its size and included in this measure).

CASE A. The two RMM agents, labeled "1" and "2", attempt to shoot down the six incoming missiles.

• Example Run:
  1. INTERACTIVE REAL-TIME DEMO (Available only at UTA)
  2. MPEG version (File size: 230K)
  3. JAVA version
  
  The team of two RMM agents, when faced with a random arrangement of the attacking missles, allows, on the average, the combined tonnage of 488 to penatrate the defense.

CASE B. A mixed RMM-human team tries to destroy the attacking missiles: An RMM agent is named as "1" and a human agent as "2".

• Example Run:
  1. INTERACTIVE REAL-TIME DEMO (Available only at UTA)
  2. MPEG version (File size: 218K)
  3. JAVA version
  
  The total expected damage of a human and an RMM agent mixed team is, on the average, 652. Thus, it is less than that of 2 human agents and more than that of 2 RMM agents.

CASE C. The two human agents, labeled "1" and "2", attempt to coordinate to intercept the six incoming missiles.

• Example Run:
  1. INTERACTIVE REAL-TIME DEMO (Available only at UTA)
  2. MPEG version (File size: 340K)
  3. JAVA version
  
  The total expected damage of human-controlled coordinated defense is, on the average, 772, which is much more than that of two RMM agents. The most obvious reason for these disparities is that humans tend to depend on their intuitive strategies for coordination, and do not engage in deeper, normative, decision-theoretic reasoning.

The Antiair Defense System using RMM was created by Piotr J. Gmytrasiewicz and Sanguk Noh. This research has been sponsored by the Office of Naval Research Artificial Intelligence Program under contract N00014-95-1-0775. Here are the papers describing our approach in more detail:

• Sanguk Noh and Piotr J. Gmytrasiewicz, Multiagent Coordination in Antiair Defense: A Case Study, In Boman, M., and Velde, W., eds., Multi-Agent Rationality - Eighth European Workshop on Modeling Autonomous Agents in a Multi-Agent World, MAAMAW'97, Lecture Notes in Artificial Intelligence., pages 4-16. New York: Springer, May 1997.

• Sanguk Noh and Piotr J. Gmytrasiewicz, Agent Modeling in Antiair Defense, In Jameson, A., Paris, C., and Tasso, C., eds., Proceedings of the Sixth International Conference on User Modeling, pages 389-400. SpringerWienNewYork, June 1997. On-line Conference Proceedings of UM97 is available from http://um.org.

• Sanguk Noh and Piotr J. Gmytrasiewicz, Bayesian Belief Update in Antiair Defense, In the Workshop of Machine Learning for User Modeling of the Sixth International Conference on User Modeling, Italy, June 1997.

• Sanguk Noh and Piotr J. Gmytrasiewicz, Coordination and Belief Update in a Distributed Anti-Air Environment, In Proceedings of the 31st Hawaii International Conference on System Sciences, Vol. V, pages 142-151, Los Alamitos, CA: IEEE Computer Society, Jan 1998.
  * This paper was nominated for best paper of the Modeling Technologies and Intelligent Systems Track.
  

• Piotr J. Gmytrasiewicz, Sanguk Noh, and Tad Kellog, Bayesian Update of Recursive Agent Models, An International Journal of User Modeling and User-Adapted Interaction, Volume 8, Issue 1/2, pages 49-69, Kluwer Academic Publishers, 1998.

For more information, please email us.

Return to Home