Skip to content

kenneally15/BANE

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

45 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

BANE – AI War-Game Debrief System

BANE is an AI-powered war game debrief and training system that analyzes mission gameplay and user perception data to deliver personalized, Weapons School-grade feedback for Airmen across all experience levels


Table of Contents

  1. Why BANE
  2. How It Works
  3. Demo Video
  4. Key Features
  5. Roadmap
  6. Quick Start

Why BANE

  • Less than 5 % of the USAF are rated officers; yet every Airman must grasp the "family business" of air-power.
  • Great-Power Competition demands rapid, scalable tactical learning across the force.
  • Traditional human-only debrief pipelines can’t meet demand—BANE scales Weapons School expertise to everyone.

How It Works

Step Action
1 › Drag & Drop Scenario (PDF) – Upload mission rules, assets, threats
2 › Fly / Simulate Mission – Solo or instructor-led playthrough
3 › Upload Log (JSON) – Drop generated gameplay log
4 › Instant AI Debrief – Timeline, causal analysis, graded focus points
5 › Perception Analysis – Optional eye-tracking stream pinpoints what the trainee saw vs. missed

Demo Video

Scenario: INDOPACOM strike-package escort versus adversary naval group.

https://drive.google.com/file/d/1SiPJ3YWXkQwKLGsfjZ4HTV-lc09_y5gc/view?usp=sharing

https://cerebralvalley.pixieset.com/nationalsecurityhackathon/demovideos/


Key Features

Capability Description
Event Extraction Parses mission logs into an ordered timeline of tactical events
Causal Inference Links actions to outcomes using doctrinal rules and LLM reasoning
Eye-Tracking Fusion Separates decision errors from perception gaps

Roadmap

  • RL Curriculum Learning – Self-play data (à la KataGo) to evolve AI adversaries & coaching agents
  • Deep Causal Graphs – Multi-layer reasoning across mission, perception, and comms
  • Personalized Mini-Games – Micro-drills tuned to individual weaknesses
  • Additional Sensors – Heart-rate, G-force, EEG for cognitive-load analysis

Quick Start

 # Requires Python 3.13
 python3 --version
 python3 -m venv myenv
 source myenv/bin/activate
 pip install -r backend/requirements.txt
 
 # Launch back-end
 cd backend
 export ANTHROPIC_API_KEY=Your Key
 uvicorn main:app --reload

 # Launch local web front-end
 npm install
 npm run dev

Open browser and go to http://localhost:5173/
Drag in an instructor PDF (Wargame Scenario Considerations.pdf)
Fly the mission, and drop mission log JSON (scenario1.json) for instant debrief.


“BANE turns mission & perception data into Weapons School-grade debriefs today—and will fuel reinforcement-learning curricula for even smarter training tomorrow.”

About

Stanford National Security Hackathon SF - Top 6 Finalist

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 3

  •  
  •  
  •