Opportunity
SBIR / STTR #A214-42
Army SBIR Solicitation for Synthetic Sensor Data Generation Tool for AI/ML
Buyer
Army SBIR | xTech
Posted
October 26, 2021
Respond By
November 30, 2021
Identifier
A214-42
NAICS
541511, 541715
The Army SBIR program is seeking innovative solutions for synthetic sensor data generation to enhance AI/ML model development across secure environments. - Government Buyer: - U.S. Army, Small Business Innovation Research (SBIR) Program - OEMs and Vendors Referenced: - No specific OEMs or vendors named - Sensor and satellite providers referenced: World View, Digital Globe, Blacksky, RADARSAT, Capella - Products/Services Requested: - Development of a software prototype tool for generating synthetic, labeled sensor data - Targeted sensor modalities include: - Commercial Satellite/Electro Optical (World View 1,2,3, Digital Globe, Blacksky) - Synthetic Aperture Radar (RADARSAT, Capella) - Full Motion Video (FMV) - Electronic Intelligence (ELINT) - Variable Message Format (VMF) - Chat - Tool must support AI/ML model training, testing, and validation across multiple security domains (e.g., SIPR, JWICS) - Unique/Notable Requirements: - Direct to Phase II SBIR effort - Solution must generate mission-focused, labeled synthetic data for a variety of sensor types - Integration and demonstration of the software prototype required - Addresses the challenge of limited access to large-scale, labeled sensor data for AI/ML - Funding up to $1.7 million, period of performance up to 24 months
Description
The US Army requires a synthetic data generation tool for sensors such as radar to augment limited labeled training data and support AI/ML model development across multiple security domains. The project aims to develop a software prototype that can generate synthetic sensor data to improve machine learning model performance and support mission-focused synthetic data creation for various sensor modalities. The effort is a Direct to Phase II project with a duration of up to 24 months and a funding amount up to $1.7 million. The project includes phases for prototype development, demonstration, and integration of AI/ML algorithms to enhance sensor data modeling and target identification.