|
Real-Time Adversarial Intelligence & Decision-making (RAID)
Problem Statement and Objectives
RAID is a DARPA project intended to fill a critical need: automated tools that allow small unit commanders to perform predictive analysis of enemy behavior, actions, and intentions. Tested on a simulated urban battlefield, the RAID tools ingest available information regarding enemy unit order-of-battle, mission objectives, and last known location for the subset that has been detected. Using this limited information, RAID generates a prediction of the full enemy laydown, looking up to two hours into the future. Additionally, using the enemy predictions as a basis, RAID generates a recommended course of action for the BLUFOR to best counter the expected enemy activities.
Results and Benefits
The multi-year project is evaluated against a series of gate metrics to measure yearly progress. ARA's role is to generate an experimental design, specify data collection requirements, analyze the collected data, and generate an analytic report of the experimental results measured against the gate metrics. ARA innovations achieved on the RAID program include a multi-factor Combat Effectiveness Scoring algorithm and a two-pronged approach to objectively measure Prediction Accuracy. The Combat Effectiveness Scoring algorithm provides a real-time combat score that is the weighted sum of seven individual scoring factors. The two-pronged approach to prediction accuracy measurement generates both a true mathematical metric and a tactically meaningful measure in meters for the accuracy of the predicted enemy locations/strength as compared to the ground truth from the simulation.
Additional information is available from DARPA.
Primary Point of Contact:
Central Florida Division
407-823-9121
|
