Summaries - Office of Research & Innovation
Back Leverage AI to Learn, Optimize, and Wargame (LAILOW) for Strategic Laydown and Dispersal (SLD) of the USN Operating Forces
|Division||Research & Sponsored Programs|
|Department||Naval Research Program|
Kendall, Walter A.
Schwamm, Noboru E.
MacKinnon, Douglas J.
|Sponsor||NPS Naval Research Program (Navy)|
The SECNAV disperses Navy forces in a deliberate manner to support DoD guidance, policy and budget. The current SLD process is labor intensive, takes too long, and needs AI. The research questions are:
- How does the Navy weight competing demands for naval forces between the CCMDs to determine an optimal dispersal of operating forces?
- How does the Navy optimize force laydown to maximize force development (Fd) and force generation (Fg) efficiency?
We propose LAILOW to address the questions. LAILOW was derived from the ONR funded project and focuses on deep analytics of machine learning, optimization, and wargame.
Learn: When there are data, data mining, machine learning, and predictive algorithms are used to analyze data. Historical Phased Force Deployment Data (TPFDDs) and SLD Report Cards data among others, one can learn patterns of what decisions were made and how they are executed with in the past.
Optimize: Patterns from learn are used to optimize future SLD plans. A SLD plan may include how many homeports, home bases, hubs, and shore posture locations (Fd) and staffs (Fg). The optimization can be overwhelming. LAILOW uses integrated Soar reinforcement learning (Soar-RL) and coevolutionary algorithms. Soar-RL maps a total SLD plan to individual ones used in excursion modeling and what if analysis.
Wargame: There might be no or rare data for new warfighting requirements and capabilities. This motivates wargame simulations. A SLD plan can include state variables or problems (e.g., future global and theater posture, threat characteristics), which is only observed, sensed, and cannot be changed. Control variables are solutions (e.g., a SLD plan). LAILOW sets up a wargame between state and control variables. Problems and solutions coevolve based on evolutionary principles of selection, mutation, and crossover.
|Publications||Publications, theses (not shown) and data repositories will be added to the portal record when information is available in FAIRS and brought back to the portal|