I’ve been working on this project since about 2003 (you could reasonably argue that I actually started development in 1985 when I began work on UMS: The Universal Military Simulator) and I’m finally in a position to share some of this work with the world at large. TIGER (for Tactical Inference GenERator) was my doctoral research project and it was funded, in part, by a DARPA (Defense Advanced Research Project Agency) ‘seedling-grant’.
After I received my doctorate, DARPA funded my research on computational military reasoning. Since DARPA was already funding a project called TIGR (Tactical Ground Reporting System) my TIGER was renamed MATE.
MATE was created to quickly arrive at an optimal tactical solution for any scenario (battlefield snapshot) that is presented to the program. It also designed to facilitate quickly entering unit data (such as location, strength, morale, etc.) via a point and click graphical user interface. (GUI). There are three main sections to MATE’s decision making process:
- Analysis of the terrain and opposing forces (REDFOR and BLUFOR) location on the terrain. This includes the ability to recognize certain important tactical situations such as ‘anchored’ or ‘unanchored flanks’, ‘interior lines of communication’, ‘restricted avenues of attack’, ‘restricted avenues of retreat’ and the slope of attack.
- Ability to implement the five canonical offensive maneuvers: turning maneuver, envelopment, infiltration, penetration and frontal assault. This includes the ability to determine flanking goals, objectives and optimal route planning (including avoiding enemy line of sight and enemy fire).
- Unsupervised Machine Learning which allows MATE to classify the current tactical situation within the context of previously observed situations (including historically critiqued battles).
I wished to test MATE with an actual tactical situation that occurred recently in either Afghanistan or Iraq. Even though my research was supported by DARPA I did not have access to recent ‘after action’ reports. However, when I saw the HBO documentary, “The Battle for Marjah,” I realized that enough information was presented to test MATE.
The clip, below, from the HBO documentary, shows the tactical situation faced by Bravo Company, 1/6 Marines February 13, 2010:
It took only a few seconds to enter RED and BLUE unit locations to MATE (the map was downloaded from Google Earth):
Screen capture of MATE showing the Battle for Marjah tactical situation. Click on image to see full-size.
After clicking on the ‘Calculate AI’ icon, the ‘Analyze and Classify Current Situation’ button and the, ‘Generate HTML Output and Launch Browser’ button, MATE’s analysis of the tactical situation was displayed. Total elapsed time was less than 10 seconds (on a Windows XP system, or 5 seconds on a Windows 7 system).
MATE then automatically generated HTML pages of its recommendations including graphically displaying optimal paths for an envelopment maneuver that encircled enemy positions:
MATE output for Envelopment Maneuver COA. Click on image to see full-size.
MATE automatically produced HTML pages of its analysis and optimal course of action (COA) routes and instructions and launched the default browser on the computer.
To see the actual HTML output of MATE’s analysis of, “The Battle for Marjah” situation click here (opens in a new window).
For more information about MATE contact sidran [at] RiverviewAI.com.