Conflict is human behavior. Cutting edge human behavior models provide military simulations improved fidelity, flexibility, ease of use, and cost savings. MOVES researchers and military graduate students work in teams to bring artificial intelligence, cognitive/perceptual modeling, and emerging social simulation techniques to bear on military simulation problems. MOVES behavior simulation research has impacted a broad range of both constructive and virtual defense simulations, from training and analysis workhorses to cutting edge prototypes spanning a range of individual combatant behavior to the social and cultural behavior of populations.
Simulation Operator Workload Reduction
MOVES Master’s student LT Justin Ross, USN worked with the Navy Warfare Development Command to reduce the cost of Fleet Synthetic Training exercises. These exercises use a lot of manpower to control fixed and rotory wing platforms that drop sonobuoys to locate subs. A single operator is responsible for multiple platforms. This makes sub prosecution difficult and error prone, since while the operator is focused on one of his platforms, another may have picked up a sub track. If the operator does not take action quickly enough, the sub track can be lost. LT Ross’ thesis project produced a prototype, shown running the the JSAF Sea Combat Commander Display at right, which automated sonobuoy placement to maintain contact with sub tracks.
Cutting-Edge Applications of AI to Simulation
A MOVES team directed by Imre Balogh worked with the Marine Corps Combat Development Command to incorporate AI planning technology to control entities in the Combat XXI simulation, a staple of the Marine analytic simulation community. Describing unit behaviors in terms of hierarchical task networks, as in the diagram at right, is a substantial improvement over ordinary computer code. HTN behaviors are easier to create, to understand, and to scale to large scenarios. Other MOVES teams have worked with the Army’s TRAC Monterey to apply other AI technologies such as reinforcement learning and Monte Carlo game tree search to improve Combat XXI behaviors.
Social and Cultural Simulation
Under the direction of Steve Hall and Jeff Appleget, the FOCUS (Flow of Communication upon Society) architecture was developed with the Joint Warfare Analysis Center. An instantiation of this model, a detail of which appears at right, models broadcast and social media infrastructure and the effects of its presence or absence on sentiment across the social network of the general population. Individual members of the population process both the information in the media, as well as the promised and actual delivery of products and services, all of which impact their loyalty to various factions competing to govern them.
Cognitive and Perceptual Modeling
In a succession of projects with TRAC Monterey and the Office of Naval Research, a MOVES team has taken a cognitive and perceptual modeling approach to understanding how the information available to an individual entity in a simulation can be integrated into a semblance of human situation awareness and ability to predict the consequences of various courses of action. The BASE-IT project in the figure at right, incorporated prior knowledge together with the complete sensing history of the entities to produce a map of where invisible threats would be most likely found. This threat map can then be used to make decisions relating to search, security, and movement technique.
POC: Chris Darken
|Civil Affairs Officer (38-Gulf) Training||2014||US Army Special Operations|
|Mission Command Analysis Using Monte Carlo Tree Search, Part II||2013||U.S. Army TRADOC Analysis Center|
|COMBATXXI Behavior Development and Technical Support||2013||MCCDC|
|Understanding Optimal Decision Making||2013||Army Research Office|
|JWAC FY12 Agent-Based Modeling||2012||JWAC|
|P3 Puckster Partial Automation||2011||NWDC|
|Coast Guard Deterrence Effectiveness Analysis||2011||Coast Guard|