Air Force Research Laboratory

Transcription

1 Air Force Research Laboratory Remote Sensing 2 November 2017 Integrity Service Excellence Stacie E. Williams, Ph.D. Program Officer Air Force Office of Scientific Research

2 Remote Sensing Program Research Areas (Sub-Tasks) OBJECTIVE: This portfolio advances the state of the art in remote sensing with a focus on Space Situational Awareness (SSA) and Combat Identification (CID) by sponsoring fundamental research to discover solutions for real world USAF challenges. Beam Propagation and Control: Understanding the mathematical underpinnings of atmospheric turbulence for prediction and mitigation. Turbulence affects remote sensing and beam control Traditional Reconstruction AFOSR Advancement SSA and Astrodynamics: Improving monitoring and orbit prediction of space objects through better detection, tracking, characterization and dynamic modeling techniques. Imaging: Seeking novel imaging and image reconstruction techniques to overcome challenges of imaging at stand off distances. 2

3 Remote Sensing Vision Twenty Year Vision: The USAF has a contiguous surveillance architecture from ground to space that persistently and rapidly detects all threats regardless of type, range, or signal size. Space has historically has been a pretty benign domain. it s become much more contested and congested -- Gen. John W. Raymond, July 2017 Our adversaries are changing the space environment in which military space forces operate from a peaceful operating domain into something much more challenging. It's gone from just worry about tracking and cataloging, just focus on safe flight, focus on mutual cooperation, and instead now it's threat-based, denied access, or worse, it's even a denied flight environment. That's the kind of environment we're having to plan for and deal with today. Gen. John E. Hyten, Commander, Air Force Space Command, August

4 Current Priorities Next-Generation Sensors Go beyond imaging and tracking Near-exquisite threat characterization Persistent Monitoring Extend characterization to small and/or far range threats Advancing unresolved techniques Temporal Photometry Routine Daylight LEO Imaging ISAL Sim Image Range Independent 24 Hour Characterization Making Sense From Unresolved Imagery Rapid Data to Inference Data fusion Smart tasking Rapid characterization Astrodynamics Improving orbit determination Understanding the physics of satellite orbits Improving Decisional Information Previous Model New Model Understanding Physics of Orbits 4

5 Portfolio Strategy Identify USAF S&T Challenges Strategic guidance documents Participate in AF SSA S&T needs process SMC/SYG, AFSPC, NASIC AFRL/RD and AFRL/RV Stay connected to AF Operational Community NSDC, JSpOC Collaborate with the intelligence community NRO, DIA, IARPA Traditional Reconstruction AFOSR Advancement 5

6 Portfolio Strategy (ctd) Identify Potential Solutions to AF S&T Challenges Maintain currency with relevant R&D efforts Foster 6.1 to 6.2 transition by facilitating collaborations with academic PIs and AFRL/RD, RI, RV, RY Bring together academicians and user community to provide solutions to S&T challenges to include NASIC, SMC, AFSPC, NRO, DIA, FAA JASON study sponsor 6

7 Internal and External Collaborations S&T Needs Process Remote Sensing Active Collaborations In Progress 7

8 Basic Research Challenges/ Program Concerns Improving basic research transition Working with AFRL/RV to develop transition strategy We are losing the space race; we need to take more risks Need a larger investment in basic research for remote sensing Identifying the best researchers to solve important AF needs University site visits Conference and workshop attendance Facilitating collaborative working group meetings to include academia and government agencies Collaborations with other R&D organizations to include DARPA, IARPA, Los Alamos National Laboratory, Naval Research Office, and Army Research Office 8

9 Rapid Data to Inference and Astrodynamics Current Efforts Volume and heterogeneity of data poses processing challenges Need to remove basic research challenges in a data to inference architecture that operates at the speed of war Currently funding 5 grants and 1 lab tasks to address these challenges 9

10 Advanced Orbit Prediction through Physics-based Learning Background: Improve the accuracy of physics based trajectories of resident space objects. Objective(s): Design a computational framework by integrating physics based approaches with machine learning. Impact to AF: - Higher accuracy for orbit prediction for resident space objects - Advance the state of the art in learning theory Funding and Duration: YIP: $332K 36 months ( ) Collaborator(s) AFRL/RD: Paul Schumacher AFRL/RV: Ryan Weisman PI: Prof. Xiaoli Bai, Rutgers, The State University of New Jersey Accomplishments & Status: Previous Model Machine Learning approach provides higher precision prediction model. Successfully developed a methodology to predict resident space objects (RSOs)' trajectories with higher accuracy than that of the current methods. Discovered that the information of the area-to-mass ratio (AMR) can be recovered from the consistent error between two estimated states in the historical data of a catalog Five Journal publications this year New Model 10

11 Optimal Sensor Tasking Through Deep Reinforcement Learning for Space Situational Awareness Background: Develop optimal tasking algorithms to enable an intelligent system for sensor planning and coordination for tracking active satellites Objective(s): Develop a mathematical framework to use deep reinforcement for SSA and sensor tasking Impact to AF: - Intelligent sensor tasking to improve SSA - Paves the way for automated sensor network for SSN Funding and Duration: YIP: $450K 36 months ( ) Collaborator(s) AFRL/RD: Paul Schumacher and Paul Kervin (AFRL DOTS prog) PI: Prof. Richard Linares, University of Minnesota Transfer learning to train a generic deep neural network for feature extraction and shallow neural network for specific tasks Accomplishments & Status: YIP awarded FY18 Develop simulation environment for investigating deep reinforcement learning for SSA sensor tasking. Realistic orbit propagator and sensor models Light curve models for unresolved space objects 11

12 Next Generation Sensors and Persistent Monitoring Characterizing remotely presents many challenges including atmospheric affects and signal collection at long stand off distances Currently funding basic research in atmospheric physics, imaging, and novel characterization Funding 17 grants and3 lab tasks to address these challenges Current GEO Image Future GEO Image 12

13 Background: Develop a rigorous theoretical and mathematical basis for understanding deep turbulence Objective(s): Develop new approaches and optical system designs for mitigating atm degradation Impact to AF: - Improves effectiveness of directed energy weapons - Improves SSA for detection and characterization Funding and Duration: MURI: $6149K 60 months ( ) Collaborator(s) AFRL/RD: Rao Gudimetla and Mark Spencer PI: Prof. Mikhail Voronstov, University of Dayton (MURI 12) Wave Optics of Deep Atmospheric Turbulence: From Underlying Physics towards Predictive Modeling, Mitigation, and Exploitation h (km) Merging atmospheric computational fluid dynamics and wave-optics Target Turbulenceinduced scintillations 0 Mauna Loa Haleakala L (km) Computational fluid dynamics simulations of atmospheric refraction and turbulence impact on directed energy, imaging, and power beaming systems UD MURI Team Accomplishments & Status: Developed theoretical framework and numerical simulation tools for atm affects over long distances Developed successful atm mitigation techniques including control algorithms >100 peer-reviewed publications Joint effects of turbulence and refractivity on long-range (149 km) laser beam propagation: impact on aim pointing accuracy & efficiency of HEL directed energy systems Laser transmitter dn/dh 10-8 (m -1 ) 13

14 Advanced Studies of Atmospheric Polarization Background: Polarization carries valuable information if we can predict and understand natural polarization Objective(s): Develop a fundamental understanding of angle of polarization and spectral influences Impact to AF: - Enable beam correction for directed energy weapons - Improve imaging under scattering conditions Funding and Duration: Grant: $1001K 36 months ( ) Collaborator(s) AFOSR: Stacie Williams (YIP and Grant); PI: Dr. Joseph Shaw, Montana State University Accomplishments & Status: Model of skylight polarization for nm New method for determining cloud phase (liquid/ice) using all-sky polarization imaging All-sky polarization images recorded during solar eclipse Published book: Optics in the Air Observing optical phenomena through airplane windows (SPIE 2017) All sky images of degree of linear polarization before (left) and after (right) total eclipse. Cloud phase images as a function of scattering media. From top to bottom: liquid clouds, ice clouds, liquid and ice clouds Solar Eclipse Expedition 21 Aug Rexburg, ID 14

15 Twenty-four Hour, Horizon-to-Horizon Imaging with the AEOS and STARFIRE Telescopes Background: Achieve diffraction-limited resolution for observations acquired through strong turbulence with meter class telescopes Objective(s): Use numerical simulation along with Monte Carlo techniques and real data to develop novel imagery processing methods Impact to AF: - Enables 24 hour imagery - Provides improves resolution for AF telescopes Funding and Duration: Grant: $1205K 36 months ( ) Collaborator(s) AFRL/RDS: Ryan Swindle and Mala Mateen PI: Dr. Stuart Jefferies, University of Hawaii D/r 0 =70 Restoration of images acquired through strong turbulence requires good wave front estimation Accomplishments & Status: High-resolution wave front sensing through strong turbulence Diffraction-limited resolution through strong turbulence (D/r 0 =70) Proof-of-concept for a laser guide star (LGS) during the daytime Atmospheric tomography using a single LGS (for LEO targets) Dynamic aperture diversity for optimal data acquisition configuration New optimization method that is less sensitive to local minima 4 Journal publications 15

16 Visible/Thermal Signature Fusion for Improved Satellite Spin State Estimation Background: Discover improved method for identifying and characterization closely spaced objects in earth orbit Objective(s): Develop a new theory to estimate attitudes and shape retrieval using multiband information Impact to AF: - Provides status and intent of satellites - Identifies inspector satellites Funding and Duration: Lab Task: $1050K 36 months ( ) Collaborator(s) AFOSR: Carolin Frueh (new grant) AEOS Vis + LWIR Hitomi signature Spin state of Hitomi recovered for international collaboration with JAXA Accomplishments & Status: Generalized spin state estimation technique to incorporate both visible and LWIR observations. The result is that spin state can be reliably determined with fewer observations Spin state validated against resolved imagery Phase I tools transitioned to operational use PI: Mr. Paul Kervin, AFRL/RDSM Spin state recovered with no knowledge of object RA 126.8, Dec 71.7, σ =

17 AFOSR Transition Success for SSA Operational Community Prior State of the Art for Daytime Imaging AFOSR Developed New Capability on AMOS *Presented to SECAF AFOSR transitioned program provides only resolved imagery of rapidly changing target for operational SSA. Improves Imagery of Highest Priority JSpOC Tasked Objects!!! 17

18 Summary Working to improve transitions and relevance 24/7 Imaging/detection including dim objects 10 grants 1 Lab Task Astrodynamics 5 grants 1 Lab Task Affordable space based sensors Newly awarded Lab Task Proposals under review Increasing investment in this area Automated data to inference Nachman and Riecken collaboration New YIP (Richard Linares) 18