TECH FOCUS AREAS: General Warfighting Requirements (GWR) TECHNOLOGY AREAS: Space Platform OBJECTIVE: This topic's objective is to develop machine learning (ML)-based analytic approaches and methods for autonomous detection and tracking of endo-atmospheric moving targets observed by EO/IR imaging sensors on LEO satellites. DESCRIPTION: The DoD's intelligence, surveillance, and reconnaissance (ISR) enterprise seeks to bring to bear the tactical and strategic assets needed to detect, track, and target threats posed by potential adversaries. The development and integration of space-based, tactical ISR-enabling capabilities into a highly proliferated, hybrid space architecture is one critical element of employing the ISR enterprise for all-domain tactical operations. These capabilities are fundamental to ensuring that the ISR enterprise has timely and fully continuous tracking of tactical threats on a global scale in order to make warfighting decisions. One of the challenges to persistent ISR from space is having the sensing capabilities needed to collect and generate highly accurate indications of moving targets in order to convey timely actionable information across the integrated battlespace. In order to address this challenge and move beyond the current state-of-the art in detection, identification, and classification of mostly ground stationary targets, what is needed are transformative and disruptive technologies to outperform and re-conceive the function and operations of traditional space-based sensing systems and ground-based data processing, exploitation, and dissemination (PED) systems. One such transformative technology is autonomous space-based sensing for which this research topic seeks innovative autonomous data analytics that will enable highly agile sensing systems to create and deliver moving target information (MTI) as part of autonomous space architectures. The overarching goal and desired end state of this topic is an autonomous machine learning (ML)-based analytics architecture for autonomously detecting and tracking airborne moving tactical targets using MTI data that is derived from satellite EO/IR imagery and enables automatic and adaptive messaging and tasking of multi-domain assets. The objective of this research effort is to develop machine learning (ML)-based analytic approaches and computational methods for autonomous detection and tracking of endo-atmospheric moving targets observed by EO/IR imaging sensors on low earth orbit (LEO) satellites. This research effort specifically seeks to develop autonomous methods applicable to the generation of moving target information (MTI) for target flight profiles of varying altitude ranges and durations, including in-flight airborne vehicles. The technology to be developed should focus on the need for innovative deep learning and other advanced ML methods that are not only automated and adaptive for surveillance of airborne tactical targets-of-interest from space, but also provide accurate and timely moving target information in the absence of large-scale data sets for model and algorithm training. Image simulation techniques that generate realistic training and test datasets containing moving target information are of interest, including data sets that are fully synthetic as well as those that are derived from real-world data. In addition, the research should focus on satellite imagery analytic capabilities needed for robust on-board and/or cloud-based autonomous sensing, including appropriate key performance parameters and metrics for evaluating the ability to correctly determine actual and predicted moving target information. Autonomous on-board analytics are of particular interest due to emerging data-intensive space-based sensing concepts and the need for real-time MTI. ML approaches are also sought for generating/sharing moving target information to defense messaging and tasking systems that are local/distributed, including at the edge, as part of autonomous sensing grids. PHASE I: Companies selected for Phase I will conduct a review and assessment of candidate ML-based analytic approaches and computational methods for autonomously processing large amounts of space-based EO/IR imagery for moving target information (MTI). Investigate the feasibility of potential image simulation techniques and models for creating training and test data sets for autonomous data analytics. Efforts will evaluate the challenges of real-time implementations of autonomous analytics on spacecraft processors. PHASE II: Phase II efforts will design, develop, and implement a prototype autonomous analytics architecture for generating moving target information (MTI) using ML-based and other advanced data processing methods. They will demonstrate prototype architecture's autonomous functionality and operation using synthetically created data sets of LEO EO/IR imagery of in-flight aircraft and other militarily relevant targets over full flight profiles and trajectories. Additionally, demonstrations will be conducted on spacecraft processors, cloud-based platforms, and/or PC-platforms and assess trade-offs for different computational hardware with respect to MTI accuracy and latency. The demonstration's performance of the emulated autonomous end-to-end pipeline from data collection to MTI generation to MTI message preparation will be evaluated. PHASE III DUAL USE APPLICATIONS: Phase III efforts would involve enhanced performance capabilities of the prototype autonomous analytics architecture implementation. They will demonstrate autonomous sensing capabilities as part of military exercises and other representative operational environments. Working with transition partners, they will identify and evaluate opportunities for implementation/integration in DoD and/or civilian applications requiring timely data for situational awareness. NOTES: The technology within this topic is restricted under the International Traffic in Arms Regulation (ITAR), 22 CFR Parts 120-130, which controls the export and import of defense-related material and services, including export of sensitive technical data, or the Export Administration Regulation (EAR), 15 CFR Parts 730-774, which controls dual use items. Offerors must disclose any proposed use of foreign nationals (FNs), their country(ies) of origin, the type of visa or work permit possessed, and the proposed tasks intended for accomplishment by the FN(s) in accordance with section 5.4.c.(8) of the Announcement and within the AF Component-specific instructions. Offerors are advised foreign nationals proposed to perform on this topic may be restricted due to the technical data under US Export Control Laws. Please direct questions to the Air Force SBIR/STTR Help Desk: usaf.team@afsbirsttr.us REFERENCES: 1. A.P. Williams and P.D. Scharre, eds, Autonomous Systems: Issues for Defence Policymakers, published by NATO Communications and Information Agency, The Hague, Netherlands, 2015.; 2. A. d'Acremont, R. Fablet, A. Baussard, and G. Quin, CNN-Based Target Recognition and Identification for Infrared Imaging in Defense Systems, Sensors 2019, 19, 2040. ; 3. R. Sherwood, S. Chien, D. Tran, B. Cichy, R. Castano, A. Davies, and G. Rabideau, Autonomous Science Agents and Sensor Webs: EO-1 and Beyond, IEEEAC paper 1628, Version 3, updated 20 Dec 2005. KEYWORDS: autonomous sensing (from space); machine learning analytics; EO/IR imagery; moving target information
