Army%20Digitization%20Research%20Initiative

1
Army Digitization Research Initiative

• Dr. Richard A. Volz (Computer Science)
• Dr. Tom Ioerger (Computer Science)
• Dr. John Yen (Computer Science)
• Dr. James Wall (TCAT)
• Randy Elms (TCAT)
• Look College of Engineering
• Town Hall Meeting
• May 11, 2000

2
Background

• UT Austin Texas AM proposed a joint 5 year
  congressional initiative to support digitization
  research at Fort Hood
• Congressionally funded, equally split between UT
  AM
• Army Digitization Office (ADO) solicited
  proposals from Army Major Commands
• Major Commands and Agencies Participating
• National Simulation Center
• OneSAF TRADOC Program Office
• Simulation, Training and Instrumentation Command
• Central Technical Support Facility
• Full-time, on-site presence at Fort Hood

3
Tasks Funding
Tasks
?
FY 01
1 - Embedded Training Design Support 2 - Joint
Mapping Tool Kit 3D Enhancement 3 - Simulation
C4I - SIMCI Support 4 - One SAF/Staff Training
Project 5 - Training Operational Data
Synchronizer
2M Requested 1M Army budgeted
FY 00
2M Congress .5M Army
contribution
Tasks
1 - Conduct Battle Staff Training FEA 2 -
Correlate to Mission Essential Tasks 3 - Model a
Set of Staff Behaviors 4 - Conduct C4I -
Simulation Interface Analysis
2M Congress set aside
FY 99
4
Objective

• To develop an agent-based computational framework
  for simulating adaptive TOC (Tactical Operating
  Center) teamwork process for reducing the cost of
  training digital force
• Generate more intelligent/autonomous behavior at
  aggregate level (companies, battalions, brigades)
  for networked wargame simulations

5
What is agent-based adaptive TOC Teamwork ?

• Virtual TOC staff (I.e., software) interacts with
  the human trainees and adapts their actions to
• battlefield situations (e.g., enemy maneuver
  intent)
• battle plan, commanders intent
• trainees actions
• trainees profiles
• training objectives

6
Major Technical Barriers

• Needs a language for describing TOC teamwork
  processes, strategies, and procedures.
• Needs an agent algorithm that can react to
  dynamic changes in the environment.
• Need to simulate cooperative interactions among
  friendly units (e.g. information sharing,
  coordination).

7
Our Approach

• Developed a Task Representation Language (TRL)
  suitable for describing TOC teamwork processes
  and actions.
• Developed a integrated reactive planning and plan
  execution monitoring algorithm for simulating
  adaptive TOC teamwork.

8
Task Representation Language (TRL)

• A hierarchical task-decomposition language based
  on AI planning systems
• Separates tasks (what to do) from methods (how to
  do it)
• Captures staff procedures
• Decision points based on queries to a knowledge
  base (JESS)

9
The Software Architecture
Bn TOC Behavioral Knowledge (TRL)
TRL Parser
Generic Tasks, Methods, and Procedures
Reactive Planning Adaptive Execution
Instantiated Tasks, Methods, and Procedures
Load Knowledge Base
Puckster Interface
Bn TOC World Model (JESS)
Initial Battlefield Situation (JESS)
puckster
OTB
OTB-Agent Interface
Brigade Interface
Brigade Staff
10
Adaptive Plan Execution

• Detects changes in the environment using JESS.
• Reacts to the change by invoking the reactive
  planner.
• Actions include sending reports to Brigade and
  puckster

Reactive Planner
Adaptive Plan Execution
Instantiated Tasks, Methods, and Procedures
Puckster Interface
Bn TOC World Model (JESS)
Brigade Interface
11
Prototype Demo Architecture
12
Other Year 1 Accomplishments

• Encoded several Battalion tasks for a Movement
  to Combat scenario
• Developed an TRL parser to automate the
  translation from TRL knowledge to Java.
• Developed an agent interface to OTB for
  automating updates to the agents world
  model.
• Successfully implemented a distributed prototype
  system using Java and RMI.

13
Additional Funding Obtained

• Budget for Year 2 is increased by 25 to 2.5
  million.
• Succeeded in obtaining a DARPA MURI grant on
  theories and technologies for agent-based
  team/group training for the Air Force (4
  million, 3 yr 2 optional yr)

14
Lessons Learned

• Justify the use of new technology
• Agent-based digital force trainer can reduce
  training cost and improve the rigor of
  training.
• Educate the sponsor and the customer about the
  technology used.
• Part of the first IPR is a brief tutorial on
  intelligent agents.
• Balance innovation and practicality
• Keep asking us Why this can not be done by a
  government contractor? and Will it be ready for
  the demo?

15
Lessons Learned

• Manage their expectations, yet keep them excited.
  
• Prepare to answer Why cant you do X?
• Identify champions and recruit supporters
• especially from Aggies, with persistence
• Understand each stake holders self interests
• communicate and collaborate with UT, yet protect
  our interests.
• Teamwork is critical, but difficult

16
The End