June 24–25, 2020June 24–25, 2020 June 24–25, 2020. DISTRIBUTION A. Approved for public...

June 24–25, 2020

June 24–25, 2020

DISTRIBUTION A. Approved for public release: distribution unlimited.

June 24–25, 2020

Agenda – Wednesday, June 2410:30 am – 11:00 am Webinar Login

11:00 am – 11:45 am Welcome & DSO OverviewDr. Valerie Browning, Director, DSO

11:45 am – 1:05 pm Limits and Foundations – AI, alternative computing and sensingMr. Ted Senator, Program Manager, DSODr. Joe Altepeter, Program Manager, DSODr. Tatjana Curcic, Program Manager, DSODr. Rohith Chandrasekar, Program Manager, DSO

1:05 pm – 1:35 pm Break

1:35 pm – 2:35 pm Complex Social Systems – stabilization, deterrence, and teaming intelligenceDr. Bartlett Russell, Program Manager, DSODr. Jiangying Zhou, Program Manager, DSOLt. Col. David Lewis, USAF, Program Manager, DARPA DSO

2:35 pm – 3:35 pm Anticipating Surprise – resilient systems, and operations in harsh environmentsDr. Mark Wrobel, Program Manager, DSODr. Anne Fischer, Program Manager, DSODr. Bill Carter, Program Manager, DSO

3:35 pm – 4:05 pm Doing Business with DARPADr. Philip Root, Deputy Director, DSOMr. Scott Wenzel, Assistant Director, Program Management, DSO

4:05 pm – 4:35 pm Contracting 101/DisruptioneeringMr. Desmond Donaghue, Contracting Officer, CMO

4:35 pm – 5:10 pm Q&ADr. Valerie Browning, Director, DSO

5:10 pm Adjourn


June 24–25, 2020

Welcome & DSO Overview

Dr. Valerie BrowningDirector, DSO


4

DARPA Technical Offices

BIOLOGICALTECHNOLOGIES OFFICE

DEFENSE SCIENCES OFFICE

INFORMATIONINNOVATIONOFFICE

MICROSYSTEMSTECHNOLOGYOFFICE

TACTICAL TECHNOLOGYOFFICE

STRATEGICTECHNOLOGYOFFICE

Biology for security• Outpacing

infectious disease

• Neurotechnology

• Gene editing & synthetic biology

• Frontiers in math, computation & design

• Limits of sensing & sensors

• Complex social systems

• Anticipating surprise

• Symbiosis: partner with machines

• Analytics: understand the world

• Cyber: deter cyber attack

• Electronics: drive solutions for DoD access and infrastructure

• Spectrum: focus on usability of highly-adaptive systems

• Sensors: enablehigh-end capabilities to proliferate intothe field

Enterprise disruption: platforms, systems, and technologies that enable new warfighting constructs

Crosscutting themes• Eliminate high-

value assets

• Exploit cross-domain seams

• Enable decision-making asymmetry

• Lethality: resilient killchains over invulnerable systems

• Surprise: heterogeneity over uniformity

• Continuous speed: agility and adaptability over performance


DISTRIBUTION A. Approved for public release: distribution unlimited. 5

DARPA and the Defense Sciences Office

DARPA: Create and prevent technological surprise

DSO—”DARPA’s DARPA”• Creates opportunities from scientific discovery• Invests in multiple, often disparate, scientific disciplines--everywhere the

rest of DARPA is, and more• Focuses on mission-informed research

DSO: The Nation’s first line of defense against scientific surprise

6

Globalization and proliferation of technology (peer, non-peer, and non-state actors) implies that U.S. can no longer rely on having/keeping the technological advantage

DSO Mission Focus: Disruptive Innovation

“A more lethal force, strong alliances and partnerships, American technological innovation, and a culture of performance will generate decisive and sustained U.S. military advantages.”

Technology Opportunities• Early identification of S&T with potential to significantly disrupt

operational paradigms• S&T breakthroughs to mitigate the use of emerging technologies by

our adversaries• Increase speed of adaptation, fielding• Ability to maintain superiority in austere, remote, and contested

environments

$0

$100,000

$200,000

$300,000

$400,000

$500,000

$600,000

$700,000

0 5 10 15 20

Japan

Russia

2000

2005 2010 2015 2020

Gross Domestic Expenditure on R&D ($M)

https://stats.oecd.org/Indexaspx?DataSetCode=MSTI_PUB


7

The speed/complexity of military engagement is increasing

Technology Opportunities• New sensors and sensor modalities to rapidly incorporate

unique information• Accurate models of physical, environmental, and social effects

to support decision making across multiple domains• Alternatives to traditional computing/machine learning for

faster, more robust decisions• Approaches to impose complexity on adversary

“We face an ever more lethal and disruptive battlefield, combined across domains, and conducted at increasing speed and reach.”

DSO Mission Focus: Need for Operational Speed

Mach 10 12 min

iiss.org/publications/strategic-survey/strategic-survey-2018-the-annual-assessment-of-geopolitics/ss18-04-strategic-policy-issues-4

Mach 5 25 min

Mach 2.5 50

min

Mach 0.82 hr 30

min

5 min1.5 min3 min

10 min

20 min

18 min6 min

GUAM

1 hr


8

DSO Mission Focus: Undergoverned Spaces

U.S. options for escalation, deterrence, and stabilization are significantly complicated by the growing diversity of local and global actors with distinct interests, motivations, and values, especially in undergoverned spaces

“We are facing increased global disorder… Both revisionist powers and rogue regimes are competing across all dimensions of power… deliberately blurring the lines between civil and military goals.”

Technology Opportunities• Capabilities to predict consequences of actions by understanding and

modeling decision-making, intentions, and reactions of adversaries/allies• New options for measured DoD escalation or mitigating adversaries’

escalation• Proactively characterizing undergoverned spaces in terms of actors,

incentives, and dynamics • Sustainment of distributed, small unit operations in urban environments

https://commons.wikimedia.org/wiki/File:Syrian_Civil_War_map.svg#filehistory

Map of the Syrian Civil War


Infrastructure System Stability

9

Increase in adversaries’ capabilities and intent to pose existential threats to the U.S. and its allies

NDS calls for resilience across the investment portfolio including assured space operations, C4ISR, forward force maneuver, and logistics.

DSO Mission Focus: Resilient Systems

Technology Opportunities• WMD/WMT sensors and sensor networks that can warn early

enough to avoid and/or treat • Characterization of adversaries’ ability/intent to threaten U.S.

infrastructure• Robust/resilient systems and infrastructure (materials, food, water,

power, etc.)Norway

USA

Internal Waters

Canada

Denmark

Iceland

Russia

Norway/Russia SA

Russia/USA SA

Canada/USA EEZ

Unclaimed

Rare earth uses and sources

Competition for Resources

Food Water

Power


10

Current DSO Thrust Areas

Limits of Sensing & Sensors

Complex Social Systems

Anticipating Surprise

Frontiers in Math, Computation &

DesignThe Economist, April 2012

(quantum sensing, imaging through scattering media, novel light matter interactions, 3D scene reconstruction)

(WMD/WMT detection, materials for harsh environments, advanced manufacturing, autonomy)

(new social science tools and methodologies, human-machine teaming, wargaming, deterrence)

(quantum information processing, alternative computing, foundational AI science, design tools)

© 2007 Ned Batchelder


http://www.google.com/url?sa=i&source=images&cd=&cad=rja&uact=8&ved=0CAgQjRw&url=http://www.economist.com/node/21553006&ei=RaUmVcedL8HYsAW_5IKgAg&psig=AFQjCNF1WOSfHZbMzCCGlGyVJ4l_DHajjw&ust=1428682437849758

11

How We Think: The Heilmeier Catechism

Important questions to consider when approaching DARPA with ideas:

• What are you trying to do?

• How is it done today and who does it? What are the limitations of the present approaches?

• What is new about our approach, and why do we think it will succeed?

• If we succeed, what difference will it make?

• How long do we think it will take?

• What are our mid-term and final exams?

• How much will it cost?


12

Young Faculty Award (YFA)

Identify and engage rising stars in junior research positions, emphasizing those without prior DARPA funding, and expose them to DoD needs and DARPA’s program development process

Develop the next generation of academic scientists, engineers, and mathematicians who will focus a significant portion of their career on DoD and National Security issues


13

Disruptioneering

Disruptioneering is a DSO rapid acquisition approach to increasing the speed of innovation:• High risk concept exploration

• Acquisition tailored to speed (idea to program in 90 days)

• “Harness and protect the National Security Innovation Base”

• “Deliver performance at the speed of relevance”

– National Defense Strategy


Artificial Intelligence Exploration (AIE)

AIE will enable DARPA to fund pioneering AI research to discover new areas where R&D programs may be able to advance the state of the art

• The pace of discovery in AI science and technology is accelerating worldwide

• The AI Exploration (AIE) program is part of DARPA's broader AI investment strategy that will help ensure the U.S. maintains a technological advantage in this critical area

• Program Announcement (PA) release: August 1, 2019• https://beta.sam.gov/opp/5a5b919dcc337814fe57eb70c147bd72/view#general

14

This new approach enables DARPA to go from idea inception to exploration in fewer than 90 days!


https://beta.sam.gov/opp/5a5b919dcc337814fe57eb70c147bd72/view#general

Program Managers

Michael FiddyElectromagnetic Waves, Scattering & Structures

Jan VandenbrandeMath, Design, & Production Automation

Anne FischerChemical Systems

Bill CarterMaterials Science

Rohith ChandrasekarOptical, Infrared, & Radar Systems

Maj David LewisPhysics

Ted SenatorArtificial Intelligence

Mark WrobelRadiation Science & Health Physics

15

Tatjana CurcicQuantum Information Science

Jiangying ZhouArtificial Intelligence

Bart RussellBehavioral & Cognitive Science

Joe AltepeterQuantum, Spectrum, Visualization


16

Evolutionary vs. Revolutionary R&D

“The flying machine which will really fly might be evolved by the combined and continuous efforts of mathematicians and mechanicians in from one million to ten million years”

• The New York Times • 9 October 1903

“We started assembly today”• Orville Wright’s Diary

• 9 October 1903


17

• The chat feature has been disabled

• Please submit your questions via the Q&A button and we will respond to as many as possible

• Presentations will be posted on the DARPA website after the event

• Email questions about the BAA to [email protected]

• You will receive a survey following the event, please complete as your feedback is valuable and helps us to plan for future events

• Sign up for DSO News Updates via Constant Contact at http://www.darpa.mil/work-with-us/interact-with-DSO (Under Resources)

• Find PM bios, program information, and contact PMs at http://www.darpa.mil/about-us/offices/dso

Housekeeping


mailto:[email protected]

http://www.darpa.mil/work-with-us/interact-with-DSO

https://www.darpa.mil/about-us/offices/dso

June 24–25, 2020

M

Limits and Foundations –AI, Alternative Computing and Sensing

Program Managers, DSO

Mr. Ted Senator Dr. Joe Altepeter Dr. Tatjana Curcic

Dr. Rohith Chandrasekar


June 24–25, 2020

Mr. Ted SenatorProgram Manager, DSO


20

Continual Knowledge Loss Due to:Replacement, Immediate/Minutes to Days (shift change handoff)Rotations, Medium/Week to Months (to/from theater)Retirement, Long/Months to Years (from workforce)

A New Approach to Sharing Organizational Knowledge

Increasing Time

State-of-the-Art: “On the flow” Management incentives Document based


21

• A new generation of coordinated “In the Flow” Knowledge Management• Shift technology focus from organizing and indexing to collection and dissemination• Create expert light-weight tools that seamlessly interact with all teammates, including non-experts• Quickly capture and produce relevant experience & knowledge• Multi-media interaction via light-weight natural language dialogue, speech & visual modalities

A New Approach to Sharing Organizational Knowledge

Informed Knowledge Collection Tools Team and Task Aware Dissemination Tools

Just in time, just enough, & just for me


22

• Machine learning has numerous shortcomings• Difficult to incorporate domain knowledge• Cannot discriminate between outliers and noise• Poor extrapolation, easily spoofed• Learning of complex dynamics unacceptably slow

• Formal models can improve learning• Reduce date required• Segregate signal from noise• Create local prediction with global awareness • Incorporate underlying symmetries in architectures

(e.g., rotational)

Learning from Formal Models

Rapid adaptation constrained by domain knowledge


Transform theories into actions• Game description languages for non-expert

game specification, with explicit assumptions, counterfactual reasoning, etc.

• Real-world data driven contingencies to reduce computational complexity

• ML to project/embed game abstractions and increase sophistication of action space

• AI driven model validation• Robust opponent models to capture non-rational

play and deception mechanisms

23

Human Aware Adversarial Reasoning

Putting game theory into practical applications


24

• Adaptive systems have arrived• Reliable performance prediction

• Bound the effects of model changes• Guarantee against extreme failures• Optimize for performance over time

Engineering Methods for Adaptive SystemsComplexity

Information systems

Cyber physical systems

Adap

tive

(AI)

Non-

adap

tive

Freq

uenc

y of

cha

nge

Mission criticality

https://dtsproweb.defensetravel.osd.mil/dts-app/pubsite/all/view/

https://www.af.mil/News/Article-Display/Article/2195671/af-week-in-photos/

Assuring correct performance in adaptive systemsDISTRIBUTION A. Approved for public release: distribution unlimited.

June 24–25, 2020

Dr. Joe AltepeterProgram Manager, DSO


26

When Will We Know If Quantum Computers Are Useful?

Dual IBM 7090’s at NASA during Project Mercury

Plausible Applications

Molecular simulation, many-body physics, machine learning, optimization, and many others

What quantum computer would you need to test if a particular

application is viable?

Questions About Resource Scaling

Fault-tolerant vs NISQ?

Gate-based vs high-coherence annealers?

Photonic vs matter-based?https://en.wikipedia.org/wiki/IBM_7090


27

Economic Spectroscopy

Google maps traffic data for DC area at 5pm on May 27, 2020 Notional values for common magnetic signals and noise floors

What is the best way to get instantaneous, local data on economic activity?

https://www.google.com/maps/


28

Visualizing Complex Information

Audio Spectrogram Album Cover for Led Zeppelin IV

https://en.wikipedia.org/wiki/Spectrogram https://en.wikipedia.org/wiki/Led_Zeppelin_IV


29

Visualizing Complex Information

Rocky Mountains Appalachian Mountains

How can complex data be mapped into a medium that enables human intuition?

https://en.wikipedia.org/wiki/Rocky_Mountains https://en.wikipedia.org/wiki/Appalachian_Mountains


June 24–25, 2020

Dr. Tatjana CurcicProgram Manager, DSO


31

Optical Clock Network

Can we improve global-scale DoD timing accuracy from 10 ns to <10 ps?

Seeking ideas for…• Low SWaP and long-term performance• Sub-picosecond two-way time transfer

between moving/flying platforms under turbulence

• Integration of clocks and time transfer

DARPA investments enabling potentially portable optical clocks:• New optical clock designs: Cryogenic silicon cavity, two-photon rubidium vapor,

and iodine vapor• Time transfer: demonstrated femtosecond optical two-way time transfer in the

presence of motion


32

Applications for Noisy Intermediate-Scale Quantum (NISQ) Processors

How can we use NISQ processors that are expected to be on the cloud in 2020?

Google Sycamore IBM Q

Honeywell IonQ

Seeking ideas for…• Noise resilience techniques to enhance the

power of NISQ processors• Compiler optimization to overcome low

circuit depth• Algorithms, customized for applications

Can NISQ processors demonstrate a quantum advantage for real-world applications?

http

s://

ww

w.n

ytim

es.c

om/2

019/

10/3

0/op

inio

n/go

ogle

-qua

ntum

-co

mpu

ter-s

ycam

ore.

htm

l

http

s://

ww

w.y

outu

be.c

om/w

atch

?v=u

tfOzR

fqle

s

http

s://

spec

trum

.ieee

.org

/tech

-ta

lk/co

mpu

ting/

hard

war

e/ho

neyw

ells-

ion-

trap-

quan

tum

-com

pute

r-mak

es-b

ig-le

ap

http

s://

ww

w.h

pcw

ire.c

om/2

019/

06/0

5/qu

antu

m-u

psta

rt-io

nq-s

ets-

sight

s-on

-cha

lleng

ing-

ibm

-rige

tti-o

ther

s/


33

Analog Quantum Simulation (aQuSim)

Seeking ideas for…• Identifying problems of broad impact and

DoD relevance• Scaling up simulators to with reliable results• Validation & Verification

Can aQuSim tools be applied to complex problems of broader practical impact?

Quantum Gas Microscope images of atoms on a square lattice

A hyperbolic lattice

Markus Greiner, Harvard

Immanuel Bloch, MPQ, Germany Andrew Houck, Princeton

aQuSims• Trapped neutral atoms and ions• Superconducting qubits and resonators• Quantum dots and trapped electrons


June 24–25, 2020

Dr. Rohith ChandrasekarProgram Manager, DSO


35

Bringing Night Vision to Eyeglass Form Factor

Current night vision systems are heavy, bulky, and induce physical strain on the Warfighter.Current systems use centuries-old optics and decades-old electronics.

NV Objective

Courtesy of Dr. Sean Moore, NVESD

https://www.agmglobalvision.com/agm-nvg-50-3aw1

https://www.tripsavvy.com/best-golf-sunglasses-4169518

https://www.photonis.com/system/files/2019-03/How-an-Image-Intensifier-Tube-Works.pdf

Technologies that could enable direct vision of multiple infrared bands in an eyeglass form factor


36

Fundamental Limits in Antenna Bandwidth

Antennas are critical to several technologies that support DoD missions: Radar, EO/IR, Comms.Current systems operate with limited bandwidth and require significant power.

https://www.esa.int/About_Us/ESOC/ESA_open-source_software_supports_Germany_s_TerraSAR-X

https://spacenews.com/capella-addvalue-pact/

https://terasense.com/terahertz-technology/radio-frequency-bands/

https://www.rfwireless-world.com/Terminology/Advantages-and-Disadvantages-of-L-S-C-X-Ku-K-Ka-Frequency-Bands.html

Understanding the fundamental limits of super-broadband antennas, how they can be physically implemented, and the impacts it would have on DoD capabilities


37

Enhancing Control of Thermal Emission

Methods for robust, dynamic control of emissivity and emission could protect critical assets. This is theoretically understood, but has not yet been demonstrated experimentally at relevant scales.

Courtesy of Dr. Michelle Povinellihttps://physicsworld.com/a/spacecraft-in-extreme-environments/

Wu et. al., Optica 4, 1390-1396 (2017)

New material technologies and methodologies that could enable enhanced control of thermal emission over large surfaces and in relevant environments


June 24–25, 2020

M

Q&A


Mr. Ted Senator Dr. Joe Altepeter Dr. Tatjana Curcic

Dr. Rohith Chandrasekar


June 24–25, 2020

1:05-1:35

Break


June 24–25, 2020

The chat feature has been disabled but please submit your questions via the Q&A button and we will respond to as many as possible

Complex Social Systems – Stabilization, Deterrence, and Teaming Intelligence


Dr. Bartlett Russell

Dr. Jiangying Zhou

Lt. Col. David Lewis, USAF


June 24–25, 2020

Dr. Bartlett RussellProgram Manager, DSO


42

Human Factors in the Age of Autonomy

How can we enable operator/battlefield innovation with AI-enabled systems?

Adapted from Onnasch, L., Wickens, C. D., Li, H., & Manzey, D. (2014). Human performance consequences of stages and levels of automation: An integrated meta-analysis. Human factors, 56(3), 476-488.

Bypass the attentional bottleneck

Overcome the speed/accuracy tradeoff

Develop for complete performance, not component performance

Theory-derived, rather than bespoke, designs across domains

Inverting the relationship between increased autonomy and off-nominal performance


43

Acting to Understand in Undergoverned Spaces

Army’s Future Land Operating Concept (AC FLOC), Canberra: Army Headquarters (2009)

ASDA Loop

Quantify the interconnectivity among distinct layers in societal systems – from static geography to dynamic social networks

Develop “act to sense” tools that address the illegibility of undergoverned spaces

Precise

Persistent

Non-disruptive

Permeating

Sociocartography: Understanding the human domain within contextDISTRIBUTION A. Approved for public release: distribution unlimited.

44

Predicting Divergence in AI-Enabled Systems

Convergent evolution

Divergent behaviors at the edge Can an phylogenic approach to AI help

predict…

… edge case behavior?

… emergent behavior?

… emergent behavior during competition?

Can we scale methods for…

… anticipating divergence from coding and architecture choices?

…anticipating edge case behavior in systems with recursive, ensemble, or otherwise combined

approaches? Anticipating asymmetric behaviors of non-deterministic systems


June 24–25, 2020

Dr. Jiangying ZhouProgram Manager, DSO


46

Portfolio & Interests

Competency Aware Machine Learning (CAML) Nature as Computer (NAC) Artificial Intelligence Research

Associate (AIRA)Revolutionary Enhancement of Visibility by Exploiting Active

Light-fields (REVEAL)

Enable ML to provide human partners insight to skills,

experience, and reliability for tasks

Harness complex dynamic physical phenomenon to perform computations “for free”

Develop new imaging technologies capable of non-line-of-sight (NLOS) full 3D scene reconstructions from

a single viewpoint

Elevate AI to the role of an insightful and trusted

collaborator in scientific discovery and the scientific process

Current interests include machine learning, artificial intelligence, unconventional computing, sensing, surveillance and reconnaissance (ISR) exploitation technologies


47

Essential Ingredients for Teaming Intelligence: Scale-Up to the Complexity Challenge

Key ingredients of success• Known play scripts• Online self play, adversarial playoff• Ten months of brute-force training• 770 PFlops/s•days of compute

Key challenges• Unknown play scripts• No online human models• Must optimize for the team• Resource/cost for human machine training

Creating super-AI teammates is orders-of-magnitude more complex than creating superhuman opponents

nasa.govhttps://www.youtube.com/watch?v=rFGk0R5y5mw

Develop capabilities for ML to efficiently learn to be super-teammatesDISTRIBUTION A. Approved for public release: distribution unlimited.

48

Essential Ingredients for Teaming Intelligence: Develop Temporal Intelligence

Effective human machine teaming will not be optimal without building-in the fundamental elements of temporal intelligence into AI

“Time is … an unavoidable framework of the human mind that preconditions possible experience.” – Immanuel Kant

Temporal Intelligence: The ability to sense the inherent temporal characteristics of the world, to reason with it, and/or use it to modulate behavior

Past Present FutureTime

Develop capabilities for ML to understand and utilize time characteristics of information and knowledge


49

Essential Ingredients for Teaming Intelligence:Become a Versatile Research Assistant

21st Century Science Overload>7M new scientific papers/year = 20,000

papers/dayDigital Research Assistants• Keep up-to-date with latest scientific

knowledge from publications• Eliminate unproductive findings• Identify novel/unusual discoveries• Identify gaps/unknowns in current

scientific models • Provide critique of manuscripts by

scientists• …

Develop personalized AI tools for scientists to accelerate scientific discovery


June 24–25, 2020

Lt. Col. David Lewis, USAFProgram Manager, DSO


51

The Early Physicists Who “Knew It All”

Archimedes287 BCE – 212 BCE

Galileo1564 - 1642

Kepler1571 - 1630

Newton1643 - 1727

Faraday1791 - 1867

Maxwell1831 - 1879

Hamilton1805 - 1865

*this is far from a comprehensive list

Boltzmann1844 - 1906

http

s://

en.w

ikipe

dia.

org/

wiki

/Willi

am_R

owan

_Ham

ilton

http

s://

en.w

ikipe

dia.

org/

wiki

/Jam

es_C

lerk

_Max

wel

l

http

s://

en.w

ikipe

dia.

org/

wiki

/Lud

wig

_Bol

tzm

ann

http

s://

en.w

ikipe

dia.

org/

wiki

/Mich

ael_

Fara

day

http

s://

en.w

ikipe

dia.

org/

wiki

/Arc

him

edes

http

s://

en.w

ikipe

dia.

org/

wiki

/Gal

ileo_

Galile

i

http

s://

en.w

ikipe

dia.

org/

wiki

/Joh

anne

s_Ke

pler

http

s://

en.w

ikipe

dia.

org/

wiki

/Isa

ac_N

ewto

n


52

The “Last Person Who Knew Everything”

Time

Know

ledg

e

Last person who knew everything

Person’s Capacity

θ

DATA

Thomas Young1773 - 1829

Accelerate Discovery?

Today

Conferences“love them or hate them”

Dedicated AI for Team Assimilation

Hypothesis: We can accelerate discovery if we can systematically raise the collaborative knowledge capacity of people

http

s://

ww

w.a

maz

on.c

om/L

ast-M

an-W

ho-K

new

-Ev

eryt

hing

/dp/

0131

3430

41

https://conferencemonkey.org/insight/8-benefits-of-attending-conferences-1038397


53

How Might We Pull This Off?

Imagine the DATA Collaboration Room

Exploit 2nd & 3rd Wave AI to increase knowledge and collaboration capacity of multi-disciplinary research teams

https://www.steelcase.com/

Natural Language Processing

Full text search and understanding

Sentiment Analysis

Curious/Questioning AIs

Augment Reality Interfaces

Capture the collaboration Participate in the collaboration

Speech recognition

Real-time data and database access

Word2vec contextual

understanding

PAI AIRA CAML L2MLwLL ASKE D3M SD2XAI

Monitor physiological signals

and response

Infer and pose cross-disciplinary

ideas

Develop unique models

June 24–25, 2020

The chat feature has been disabled but please submit your questions via the Q&A button and we will respond to as many as possible

Q&A


Dr. Bartlett Russell

Dr. Jiangying Zhou

Lt. Col. David Lewis, USAF


June 24–25, 2020

M

Anticipating Surprise – Resilient Systems and Operations in Harsh Environments


Dr. Mark Wrobel Dr. Anne Fischer Dr. Bill Carter


June 24–25, 2020

Dr. Mark WrobelProgram Manager, DSO


57

Nanoscale Materials and Devices for Photon Sensing

Accelerate development of high performance photon sensors based on nanoscale material systems for energies ranging from sub-eV to MeV

Zherebetskyy et al., Science 344, 1380 (2014)

www.chemicalstructure.net www.mirion.comwww.analogic.com

www.flir.com

Artificial Intelligence

Multi-scale Modeling

High Throughput Screening

Multi-objective Optimization


58

Energy & Power Sustainment in Spartan Terrains

Develop novel approaches to gathering and storing energy in remote, austere operational environments

Wang et al., ACS Nano 2019, 13, 2, 1107

Rob Felt, Georgia Tech

https://jenaxinc.com/products/batteries/

https://api.army.mil/e2/c/-images/2010/11/28/93130/army.mil-93130-2010-12-02-081209.jpg


59

Buffers for Undergoverned Spaces

Develop new options to engage friendly/non-friendly local populations while minimizing the economic and social impact of stabilization operations

https://static.dvidshub.net/media/pubs/pdf_8123.pdf

https://www.commondreams.org/views/2014/07/30/behind-door-number-three-iraq


60

Directional Neutron Sources

High yield, directional neutron sources with low size, weight, and power enabling new in-the-field applications

1 m10 m

www.pelletron.comCourtesy of LLNL


June 24–25, 2020

Dr. Anne FischerProgram Manager, DSO


62

Automation and AI Have Advanced Our Capabilities to Produce and Discover New Molecules

Coley et al., Science 365, 557 (2019)

Artificial Intelligence Automation


63

These Advances May Also Introduce Vulnerabilities

Networked analytical equipment

Genetic and digital molecular data

Remotely operated cloud labs

011010011001110100010

For example, in…

Genetic Digital


64

But, This Leads to Many Opportunities (A Few Thoughts of Many)

Secure development and utilization of automated and/or cloud-based production, development, testing

Methods to ensure quality of chemical products

Resiliency and flexibility of production methods across the chemical supply chain

What solutions does the chemistry have to offer?

What ways can we exploit electro-, photo-, mechano-, … approaches to increase thermodynamic and kinetic efficiency?

How do we bake in validation?

Molecular data quality and security assurance

How do we understand the threat space from the digital to the molecular world? What implications does this have for our future information security?


June 24–25, 2020

Dr. Bill CarterProgram Manager, DSO


Materials for Harsh Environments: Enabling the “High Ground”

Example topics of interest:• Scientific capabilities for developing extreme

materials• Risk analysis for nonlinear coupled systems• On-orbit manufacturing• Tough self-decontaminating surfaces• High-temperature metamaterials

66DISTRIBUTION A. Approved for public release: distribution unlimited.

67

Risk Analysis and Materials Science for Extreme Environments

Can we better predict vehicle response using imperfect/limited information?

https://www.nasa.gov/centers/ames/research/technology-onepagers/arcjetcomplex.html

Arc-jet test and evaluation facility• High-cost, low availability• Moderate fidelity representation of

harsh environment • Rarely used for materials development

HIFiRE

Can we develop new lab-scale techniques that better reproduce extreme conditions?

Oxyacetylene torch• Low-cost, high availability • Low fidelity representation of harsh

environment• Often used for materials development

Future need: A higher fidelity extreme

environment capability that can be hosted in a

typical lab Need new approaches to:• Risk analysis • Uncertainty quantification• Design margin allocation• Materials needs

Extreme environment platform• Nonlinear interactions • Limited test flights• Empirical models• Overdesigned

Arc-jet test

Torch test

General needs for development of extreme platformsDISTRIBUTION A. Approved for public release: distribution unlimited.

68

On-Orbit Manufacturing

Create 5-10X larger satellite structures (e.g., apertures, solar panels, sensor spacing)

Structure required to survive launch translates into ~7-14X mass penalty on orbit providing an opportunity for significant improvements in mass efficiency

Structural precision required: ∆L/L10-610-410-2

1.0

0.1

0.01

0.001

Launchcapability

Today’s deployableantennas

On-orbit manufacturin

g

Stru

ctur

al d

ensi

ty

(g/c

c)

Ideal (antenna)

Today ~5m antenna deployed

e.g., Falcon 1 faring

Future needs: On-orbit net-shape manufacturing of structural, electrical, thermal and optical materials

NASA

Future >30 m antenna

http://digitalcommons.usu.edu/cgi/viewcontent.cgi?filename=0&article=1523&context=smallsat&type=additional


69

Tough Self-Decontaminating Surfaces

Long lasting, high performance, antimicrobial multifunctional coating

Seeking broad spectrum, fast acting, wear resistant approaches


70

High Temperature Metamaterials

Combining the unique properties of metamaterials to solve high temperature materials problems

Seeking metamaterial solution that actively lower temperatures or are capable of high temperatures

Suemitsu etl al., ACS Photonics 2020, 7, 1, 80–87


June 24–25, 2020

M

Q&A


Dr. Mark Wrobel Dr. Anne Fischer Dr. Bill Carter


June 24–25, 2020

Doing Business with DARPA

Dr. Philip Root & Mr. Scott WenzelDeputy Director & ADPM, DSO


73

• Doing Business with DSO• Understanding what DSO wants

• Communicating your ideas

• Doing Business with DARPA• Understanding Broad Agency Announcements (BAAs)

• Proposing to a DSO BAA

• Important take-aways

Overview


74

Doing Business with DSO:Understanding What DSO Wants


75

DARPA Inside

People Processes

• DARPA program managers• Exceptional

technologists and researchers

• Typically serve 3-5 years

• DARPA programs• Approved by Director• ~ $45M over 5 years

• DoD, Administration, and Congress support mission and autonomy

Culture

• Drive for off-scale impact

• Risk taking • Honor in public service


76

DARPA’s Portfolio

FoundationsUnderstanding complexity, composable systems, advanced materials and electronics, trusted hardware and software, human-machine

symbiosis, 3rd wave artificial intelligence, data and social science, new computing, and engineered biologyIncreasing the pace of developing technologies and capabilities for the US and allied warfighter

Defend the homeland Effectively prosecutestabilization efforts

Deter and prevail against high-end adversary

Artist’s concept

Multi-varied threats to the nation

Continuous counter-terrorism and counter-insurgency operations

Peer competitor confrontations in Europe and Asia

Cyber deterrenceBio threat detection and mitigation

Defense against WMTCountering hypersonics

Adaptive lethality for air, land & seaControl of the EM spectrum

Long range effectsRobust space

Countering gray warfare Behavior modeling & influence

3D city-scale operationsWarrior performance


Current DSO Thrust Areas

Limits of Sensing & Sensors

Complex Social Systems

Anticipating Surprise

Frontiers in Math, Computation &

DesignThe Economist, April 2012

(quantum sensing, imaging through scattering media, novel light matter interactions, 3D scene reconstruction)

(WMD/WMT detection, materials for harsh environments, advanced manufacturing, autonomy)

(new social science tools and methodologies, human-machine teaming, wargaming, deterrence)

(quantum information processing, alternative computing, foundational AI science, design tools)

© 2007 Ned Batchelder

77DISTRIBUTION A. Approved for public release: distribution unlimited.

http://www.google.com/url?sa=i&source=images&cd=&cad=rja&uact=8&ved=0CAgQjRw&url=http://www.economist.com/node/21553006&ei=RaUmVcedL8HYsAW_5IKgAg&psig=AFQjCNF1WOSfHZbMzCCGlGyVJ4l_DHajjw&ust=1428682437849758

78

• All program ideas originate with a Program Manager• PMs identify opportunity to make a difference

• DARPA PMs typically serve 3 – 5 year terms• This means approximately 25% turnover annually

• New PMs mean new ideas and potentially investments in different research areas

Where Do DSO Program Ideas Come From?


79

Doing Business with DARPA:Communicating Your Ideas


80

• Any time prior to submission of a proposal • Via E-mails, phone calls, or face-to-face meetings • Find DSO PM bios, program information, and contact PMs at:

• https://www.darpa.mil/about-us/offices/dso

• If you have a question related to a specific BAA,contact the PM via the email address listed in that BAA

Program Managers recommend all research awarded by DARPA

When and How to Communicate with DARPA PMs



81

1. What are you trying to do?2. How is it done today and who does

it? What are the limitations of the present approaches?

3. What is new about our approach, and why do we think it will succeed?

4. If we succeed, what difference will it make?

5. How long do we think it will take? 6. What are our mid-term and final

exams? 7. How much will it cost?

What Are the Heilmeier Questions?

Also, be prepared to answer –Why DARPA? Why DSO?Why now?

FOCUS


82

What Are Steps to Get From Idea to Award?

PM discusses idea with DSO Office Director &

Deputy DirectorPM preps formal

briefing

PM presents formal briefing to DARPA

leadership

DARPA Director and Deputy decide to fund

the programBAA is published Proposals arrive

Proposals are evaluated DSO leadership approve proposals for funding Contracting


83

Doing Business with DARPA:Understanding Broad Agency Announcements (BAAs)


84

• BAAs are used when• Soliciting research and development proposals directed toward advancing the state-of-the-art

• Describing a problem or general research area(s) of interest rather than providing a common work statement

• Anticipating varying scientific/technical approaches

• Proposals submitted to DARPA BAAs are evaluated• Through scientific/technical review

• On their own merits

What Is a BAA?


85

DARPA issues two types of BAAs1. Program-specific BAAs2. Office-wide BAAs

How Does DARPA Use BAAs?

DARPA is interested in revolutionary ideas that advance DoD’s mission

Find current DARPA BAAs at:https://www.darpa.mil/work-with-us/opportunities

beta.sam.gov

https://www.grants.gov

Proposers Days are advertised at beta.sam.gov, and DSO will send out notifications via Constant Contact



http://www.darpa.mil/work-with-us/opportunities

https://beta.sam.gov/


http://www.grants.gov/

86

Doing Business with DSO:Proposing to a DSO BAA


87

• An executive summary is a brief (1 page + cover) outline of the proposed research idea• An abstract is a slightly more detailed (5 page) synopsis of the proposed research idea

• Submission of executive summaries and abstracts allows proposers to –− Quickly ascertain whether the proposed concept is of

interest to DSO

− Save bid and proposal costs

If you were not encouraged to submit a proposal based on your executive summary or abstract, you may still submit a proposal, but it is less likely that it will be funded

What Is an Executive Summary? What Is an Abstract?


88

• Follow ALL directions in the BAA• Each DARPA BAA has content/formatting/submission requirements specific to that BAA• Non-conforming proposals will not be reviewed and will not be eligible for award

• Include a straightforward, concise description of the technical solution and a fully supported cost proposal

• Ensure your proposal responds to the Heilmeier Questions

How Do I Write a Successful Full Proposal?


89

• Evaluation criteria are listed in descending order of importance• Typical criteria include:

‒ Overall Scientific and Technical Merit‒ Potential Contribution and Relevance to the DARPA Mission‒ Cost Realism

• Cost or price is rarely, if ever, the deciding selection factor• Proposals contain unique solutions - they are not compared one to another

What Are the Evaluation Criteria and Relative Importance?


90

• Unclassified submission portals• Contracts and Other Transaction proposals: https://baa.darpa.mil• Grants or Cooperative Agreements proposals: www.grants.gov• All executive summaries and abstracts: https://baa.darpa.mil• Mail/Hand-carry submission: See instructions in the BAA

How Do I Send My Submission to DARPA?


https://baa.darpa.mil/

http://www.grants.gov/

https://baa.darpa.mil/

91

• Send all questions (technical, contractual, administrative) about the BAA to the email listed in the BAA

• FAQs will be posted under the BAA at http://www.darpa.mil/work-with-us/opportunities (filter by “Defense Sciences Office”)

How Can I Get Answers to Questions About the BAA?


http://www.darpa.mil/work-with-us/opportunities

92

Once proposal evaluations are complete, proposers will be notified whether or not their proposal was selected for award negotiation

• Successful proposers will be contacted by a contracting officer to begin negotiations• You may request informal feedback from the PM regarding your proposal• Proposals may be selected for partial funding• Types of awards may include grants, cooperative agreements, contracts or other

transactions• Award is subject to successful negotiation and availability of funds

What Happens Next?


93

What Are “Disruptioneering” and “Artificial Intelligence Exploration” (AIE)?

Efforts ~$5M - from opportunity posting to award in < 90 days

“Disruption Opportunity” or “AIE Opportunity” issued as a Special Notice under the

Disruptioneering Program Announcement (DARPA-PA-19-02) or the

AIE Program Announcement (DARPA-PA-18-02)

Streamlined proposals due within 30 days

Other Transactions (OT) for Prototype awards in fewer

than 60 days

Award information• Up to 2 phases, not to exceed $1M total, including any cost share‒ Phase 1 Feasibility Study (3-9 months, NTE $500K)‒ Phase 2 Proof of Concept (9-15 months)

• Streamlined templates and instructions

Anticipated benefits• Risk reduction – Targeted investments prove feasibility• Increased innovation tempo – More research faster


94

• Read the BAA and follow the instructions• Communicate with a PM• Answer the Heilmeier Questions• Start with an executive summary or abstract• Ensure proposal is “conforming”• Be concise, but detailed• Ask questions if you don’t understand

What Are the Most Important Take-Aways?


June 24–25, 2020

Q&A

Dr. Philip Root & Mr. Scott WenzelDeputy Director & ADPM, DSO


June 24–25, 2020

Contracting 101/Disruptioneering

Mr. Desmond DonaghueContracting Officer, CMO


97

• No common Statement of Work - Proposals evaluated on individual merit and relevance as it relates to the stated research goals/objectives rather than against each other

• Proposals are evaluated for strengths and weaknesses relative to the criteria published in the BAA, listed in descending order of importance:

• Overall Scientific and Technical Merit

• Potential Contribution and Relevance to the DARPA Mission

• Cost Realism

• Government reserves the right to select for award all, some (partial selection), or none of the proposals received

• Contract negotiation timelines depend on each institution/organization’s response time

Full Proposal Review-Awards Process

Government reviewers

PM recommendation

to SRONotification Program

Kick OffContracting


98

• Read the BAA carefully• Nonconforming proposals may be rejected without review

• There is no page limit to Volume II Cost Proposal• Use provided DARPA cost proposal spreadsheet (posted with BAA)• Cost break downs are outlined in the BAA:

o By Phase: Phase I (base), Phase II (option) by contractor fiscal year

o Total program cost by major task

o Projected funding required by month (only at summary level)

o Subcontractor proposals must be prepared at the same level of detail as required of prime (Subs must use DSO cost proposal spreadsheet)

Cost Proposal


99

• Streamline the award and negotiation process• Opportunity release to award in less than 90 days

• Other transaction for Prototypes only (10 USC 2371b)• Define research requirement to enable price as an eval factor

• Milestone plan provided; detailed description of research project objectives, tasks, deliverables, travel, etc.

• Establish price reasonableness during proposal evaluation

• Streamlined OT model - mandatory review w/ track changes

• All OTs have fixed payable technical milestones

Disruptioneering

Post DO SN to beta.SAM.gov

30 days

KickoffBriefs

Scientific Review Process

15 -20 days

PR10 days

Negotiation and Award Execution

30-35 days


100

• Continuously open Program Announcement (PA) – DARPA-PA-20-01• Each Opportunity announced by separate Opportunity Notice to beta.SAM.gov

• DARPA-PA-20-01-01 (Opportunity notice ID for beta.SAM.gov)• Streamlined Proposal: Technical (8 pages), price, and admin/policy volumes• Only Other Transaction (OT) for Prototype awards; Includes model OT in PA• Proposals due 30 days after Disruption Opportunity published to beta.SAM.gov• Each OT is limited to $1M over 2 phases: $1M includes cost share if req.

• Level of Effort ceiling in notice (i.e., $400K Ph1/$600K Ph2)• Milestone plan provided

• Phase 1 Feasibility Study (base) and Phase 2 Proof of Concept (option) • 3-9 months for the Phase 1 base effort • 9-15 months for the Phase 2 option effort • PoPs may vary – not anticipated to exceed 18 months

• Evaluation criteria• Overall Scientific and Technical Merit• Potential Contribution and Relevance to the DARPA Mission• Price

Disruptioneering


101

• Must serve the DOD mission:“…carry out prototype projects that are directly relevant to enhancing the mission effectiveness of military personnel and the supporting platforms, systems, components, or materials proposed to be acquired or developed by the Department of Defense, or to improvement of platforms, systems, components, or materials in use by the armed forces.”

• Prototype defined in recent OSD Memo 11/20/18:The definition of a "prototype project" in the context of an OT is as follows: a prototype project addresses a proof of concept, model, reverse engineering to address obsolescence, pilot, novel application of commercial technologies for defense purposes, agile development activity, creation, design, development, demonstration of technical or operational utility, or combinations of the foregoing. A process, including a business process, may be the subject of a prototype project.

10 USC 2371b Requirements


102

2371b(d)(1) permits DARPA’s OT authority to be used only when one of the following conditions are met:

1. There is at least one nontraditional defense contractor or nonprofit research institution participating to a significant extent in the prototype project;

2. All significant participants in the transaction other than the Federal Government are small businesses (including small businesses participating in a program described under section H. R. 2810—213 9 of the Small Business Act (15 U.S.C. 638)) or nontraditional defense contractors;

3. At least one third of the total cost of the prototype project is to be paid out of funds provided by sources other than the Federal Government; or

4. The senior procurement executive for the agency determines in writing that exceptional circumstances justify the use of a transaction that provides for innovative business arrangements or structures that would not be feasible or appropriate under a contract, or would provide an opportunity to expand the defense supply base in a manner that would not be practical or feasible under a contract.

Use of OTs

103

• As defined by 10 U.S.C. § 2302(9), a nontraditional defense contractor, with respect to a procurement or with respect to a transaction authorized under section 2371b of this title, means an entity that is not currently performing and has not performed, for at least the one-year period preceding the solicitation of sources by the Department of Defense for the procurement or transaction, any contract or subcontract for the Department of Defense that is subject to full coverage under the cost accounting standards prescribed pursuant to section 1502 of title 41 and the regulations implementing such section. To be considered as participating to a significant extent, the proposal should substantiate that the effort being performed by the nontraditional defense contractor is critical to the technical success of the project.

Definition of Nontraditional


104

Disruptioneering Goal = RAPID PROTOTYPING

• FAR, DFARS, CICA, TC&PD, Bayh-Dole do not apply• Streamlined administrative practices

• NO D&Fs for Options, D&Fs for contract type, Sub plans, Certificate of Current Cost and Pricing Data• Scope is not a concern since CICA does not apply

• Can modify OT to continue further development beyond original solicitation• Performers apply best commercial practices to achieve program goals and remain focused on achieving

key technical milestones rather than admin. tasks• Performers empowered to look for tradeoffs & drive efficiencies rather than Government• Supports doing business with non-traditional performers

Advantages of OTs for Prototypes

Prototypes can transition into Production OTs. • Successful Prototype OTs offer a streamlined method for transitioning into follow-on production without

competition • Prototype OT must have been competitively awarded and successfully completed• Solicitation documents and the Prototype OT agreement shall include notice that a follow-on

Production OT is possibleDISTRIBUTION A. Approved for public release: distribution unlimited.

June 24–25, 2020

Q&A

Mr. Desmond DonaghueContracting Officer, CMO


June 24–25, 2020

4:35-5:10

Q&A


June 24–25, 2020

Agenda – Thursday, June 25

11:00 am – 4:20 pmIndividual Sidebars by Invite Only


Date post:	17-Jul-2020
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

June 24–25, 2020June 24–25, 2020 June 24–25, 2020. DISTRIBUTION A. Approved for public...

Documents