AVATAR Case Study Research and Technology Transition Workshop July 24-25, 2012
Agenda Introduction Video AVATAR Background Example Study Technology Transition Process SENTRI/Trusted Traveler Field Test Path to Transition / Discern Science
What is BORDERS? DHS Center of Excellence Funded by S&T University Programs Awarded July 2008 Supplemental funding for specific projects Consortium of 15 universities Headquartered at University of Arizona Leverages $250m in on-going research/facilities Mission Conduct border security and immigration research Educate the next generation of scientists and personnel Transition prototypes and products to DHS customers
BORDERS Partners
DHS Stakeholders Customs & Border Protection Immigration & Customs Enforcement Transportation Security Administration Citizenship and Immigration Services
BORDERS Research Areas RA 1: Detection, Identification and Screening RA 2: Sensor Networks and Communications RA 3: Immigration Policy and Enforcement 6
About Me Second Year Doctoral Student Technology Transfer, Commercialization Entrepreneurship Ecosystems Automated Credibility Assessment Academic History BS in Systems Engineering & Entrepreneurship MBA in Marketing and Finance Work History Founder or 1 st employee variety of startups Founded technology trade association for Tucson Current founder of Startup Tucson
8 Intro to the AVATAR
Research Goals Unobtrusive non-invasive credibility assessment No sensors attached to the body Real-time, remote analysis Scalable and robust for high traffic Useful across contexts Useful across cultures
Research Questions What video, audio, and language-based features measure changes in behavior and reliably detect deception, impostership, and malicious intent? What psychophysiological measures are reliable indicators of deception? Heart rate - Pupil diameter Changes in Blood pressure - Gaze behavior Respiration - Thermal temperature Blink Rate - Galvanic skin response Which interviewing tools and techniques are most effective for identifying imposters, smugglers, terrorists, and other criminals?
Problem Complexity: There are Many Ways To Deceive Lies Fabrications Concealments Omissions Misdirection Bluffs Fakery Mimicry Tall tales White lies Deflections Evasions Equivocation Exaggerations Camouflage Strategic ambiguity Hoaxes Charades Imposters 11
Problem Complexity Rapid Screening People usually make rapid decisions about credibility (7-20 seconds) Automated Interviews Force multiplier Interview Control and Consistency Detecting which questions are most diagnostic Question set changes for every context
Problem Complexity: Overconfidence and Underachievement Humans are poor lie-detectors ~54% accuracy rate for general population Accuracy is a function of the quality of base rates Poor performance affects novices and professionals Confidence in judgment is not correlated with accuracy Affects attentiveness, verification efforts, and misallocation of resources 13
Interdisciplinary Research Psychology, Cognitive Science, Nueroscience Communication, Linguistics, Speech & Hearing Automated Deception Detection Computer Science, Information Systems Engineering (Electrical & Systems) 14
Research Approach - Multimethodological
Automated Human Screening: 30+ Years of Collaborative, Multi-disciplinary Research Psychology: Cognitive, Experimental, Social, Organizational Insider Threat Virtual Interactions Signal Processing Bioengineering Biometrics Communication Computer Science: Computer Vision, Artificial Intelligence Polygraph Information Systems Oculometrics Criminology Neuroscience Linguistics
Automated Virtual Agent for Truth Assessments in Real-Time (AVATAR) Avatar-based screening system Rapid screening technology Credibility assessment Artificial intelligence Stand-off sensor technologies Flag suspicious behavior for secondary screening Can be customized for use with: - Ports of Entry - visa processing - airports - asylum requests - detention centers - personnel screening
Non-Invasive Tools for Rapid Screening LDV THERMAL BLINK CAMERA PUPILLOMETRY EYE TRACKING VOCALIC ANALYSIS COMPUTER VISION LINGUISTICS FORCE PLATFORM 18
Sensor Integration Platform
Sensor Research
AVATAR Kiosk - Generation 1 Lab was transformed by packaging sensors into a kiosk
AVATAR Kiosk - Generation 2 First working prototype Kiosk form factor Embedded sensors Ruggedized for commercial use
AVATAR Kiosk - Generation 3 Automated Kiosk Interactive intelligent agent Embedded sensors Avatar-based interviewer Expert system Biometric reader Multilingual capacity
25 Sample Avatar Interviewers
Example Experiments: Bomb Studies Travelers packed a bag and were interviewed by an automated screening kiosk Half of participants also constructed and packed an explosive device: 26
27
28
29
30
Sample Results: Gaze Analysis Control Average = 12.43% SD = 6.46% Bomb Average = 28.52% SD = 13.67% 31
Eye Gaze: Guilty 32
Eye Gaze: Innocent 33
The Voice of a Bomb Maker Bomb Makers had 25.34% greater variation in their vocal pitch when answer this question: Has anyone given you a prohibited substance to transport through this checkpoint? BOMB MAKER INNOCENT 34
Explaining the Variation in Pitch Intonation Pitch contours reveal a steep pitch rise over time Increased pitch at the end of an utterance reflect uncertainty Pitch (Hz) Bomb Maker Innocent Time (s) 35
TECHNOLOGY TRANSITION EFFORTS
Technology Transition Philosophy Technology transition starts at the beginning The Last Research Mile (Dr. Jay Nunamaker) Find a painful problem Propose a new solution Listen carefully to the reactions Devise an approach to ease their pain Design and build a prototype Test in the lab Test in the field This is an iterative process Bring lessons learned back to lab to drive improvement
Last Mile Research Road Map Proof-of-Concept prototype Proof-of-Value Proof of self-sustaining Use (POC) (POV) (POU) Real Problem POC POV POU 38
SENTRI - Trusted Traveler Pilot Test Expedited processing for pre-approved, low-risk travelers entering U.S. on-line application background check Personal interview consists of 20 standard questions
Trusted Traveler Pilot Test DeConcini Port of Entry, Nogales AZ Dec. 9, 2011 : 8- week field trial of AVATAR for SENTRI Program Tucson Field Office provided space at SENTRI Enrollment Center in Nogales Work and meeting space provided by UA Santa Cruz County Personal interview is time-consuming for officers Takes time away from higher-level tasks Most applicants are low risk SENTRI applications are expected to increase Hiring levels expected to remain stable AVATAR acts as force multiplier to conduct interviews and segment risk 1 officer can monitor 4-8 AVATAR kiosk and receive real-time feedback Confirm the good people and move them through quickly All others sent to officer for secondary screening
Trusted Traveler Pilot Test DeConcini Port of Entry, Nogales AZ Pilot Phase 1 (Dec 2011 Jan 2012) Conducted 175 interviews (English only) Demos: Comm. Bersin and Aguilar, Pres. Luc Portelance (CBSA) Lessons learned: Rewrite English script Translate Spanish Recognize when a person speaks over AVATAR Establish a training program Pilot Phase 2: July 2012 Incorporate researcher and CBP feedback 1000 Interviews
Commission Bersin
Future Directions Examine and improve resilience to countermeasures Mental and Physical countermeasure detection Non-contact measurement of Electrodermal Activity Partnering with Night Vision Lab Expanded SENTRI role and additional applications Lowering Language Level Reciprocal Avatar Behavior / Interactive Dialog Separating Avatar and Human voice
Potential Future Applications All Trusted Traveler Programs I-94 visas TSA domestic travel Apprehension interviews Periodic investigations New hires Customize kiosks for these applications
Path to Transition Patent process started for AVATAR technologies Vocalics, Linguistics, Kinesics, Ocular, etc. Responding to SBIRs and seeking venture capital Started new company: Discern Science Corp. The science of the ability to judge well
Discern Science was founded in 2011 Discern Science The products being developed for commercial sale were sponsored by DHS and DoD Discern Science is the science of judging well August 2, 2012 46
Commercialization Efforts 2011 Invention disclosure filed with University Provisional patent filed with USPTO Discern Science Corporation formed License option with University negotiated 2012 SBIR application filed ($250K - $1M) Standoff Counter Human Deception Detection Device - SBIR A12-060 Arizona Innovation Challenge ($250) NSF I-Corps Program ($50K)
Trusted Traveler- Domestic exploding demands beginning soon Screening Personnel CBP, $250K per year DoD, $1M per year AVATAR Kiosk Costs $100K in 1 st year, $30K each subsequent year Pain Thousands Border Patrol Workload Staffing 2012 2022 Work load for the AVATAR Discern Science - Confidential and Proprietary *** Draft *** August 2, 2012 48
Value Proposition Rapid return on investment 100% - 500% ROI on large scale implementation (20 units or more) Improved security and improved utilization of human capital The AVATAR (kiosk) received very positive comments during one of Ex-Commissioner s (Alan) Bersin s visit to the Nogales enrollment center on December 13, 2011. The pilot is ongoing and preliminary indications are that there is a potential to use this technology to assist CBP in improving manpower utilization to balance the security and facilitation elements of our mission. -- Armando Goncalvez, Program Manager Trade, Tucson Field Office, U.S. Customs and Border Protection Discern Science - Confidential and Proprietary *** Draft *** August 2, 2012 49
Customer Value 5 CBP Officers cost $1.125 mil Efficiency: 1 Customers to 1 Officer 1 CBP Officer and 6 AVATAR kiosks cost $1.15 mil Yr 1 $400K After Yr 1 Efficiency: 5 Customers to 1 Officer Year 1 ROI 9% Year 2 ROI 213% Discern Science - Confidential and Proprietary *** Draft *** August 2, 2012 50
Market (First Five Years) U.S. Customs and Border Protection $115 million (sales) $34 million (services) Department of Defense $500 million (sales) $150 million (services) Transportation Security Administration $60 million (sales) $17 million (services) $675 Million in Estimate Total Product Sales Revenue $201 Million in Reoccurring Annual Service Revenue Discern Science - Confidential and Proprietary *** Draft *** August 2, 2012 51
Competitive Advantage No direct competitor exists for the AVATAR Some of the features exist in separate products Basic ATM like kiosks exists to collect information Current Credibility Assessment technology require attaching to contact and most rely on a single sensor No competitor has scientific validity behind its product Product requires more than engineering Relies on behavioral sciences Patents and exclusive licensing of all technology integrated into an AVATAR kiosk Discern Science - Confidential and Proprietary *** Draft *** August 2, 2012 52
Go to Market Strategy (First 18 Months) Complete CBP SENTRI Program Field Trial Develop Commercial Product with AIC Funding Receive Order from CBP for Five Units for SENTRI Program Obtain Additional Funding for 2013 Operating Expenses (Est. $600k) Expand SENTRI Program Product Sales and Enter CBP NEXUS Program (25 units sold) Discern Science - Confidential and Proprietary *** Draft *** August 2, 2012 53
Corporate Development Discern Science - Confidential and Proprietary *** Draft *** August 2, 2012 54
Assumptions and Risks Critical Assumptions Negative Impacts Mitigation Target Customers adopt Technology at Rated Predicted AVATAR technology development continue as projected Funding needs in year 2 are met Continued R&D and related funding through Univ. of Arizona Slow product adoption slows revenue acquisition Delays in technology development could delay initial sale, increase cost, and future revenue Without additional funding in year 2 product development and sales will slow Future products could be delayed if funding for basic research at UA were reduced Completing 2 nd Field Trial of technology now, working to ensure product adoption Extensive on-going prototype development resulting in a well scoped commercial development project Factoring of purchase orders to fund year 2 development Develop research tools and talents within Discern to perform product development research in house Discern Science - Confidential and Proprietary *** Draft *** August 2, 2012 55
Questions
SUPPORTING SLIDES Discern Science - Confidential and Proprietary *** Draft *** August 2, 2012 57
Cues & Sensors for Deception Detection Recommendation Discern Science - Confidential and Proprietary August 2, 2012 58
Proof-of-Concept Prototype Find out if the proposed solution works Discover any limitations of the system 59
Proof-of-Value Sufficient robustness and functionality to solve ONE important, painful problem Take it into the field and shepherd the users MEASURE whether they are more productive with the system than without Write down every compliment, and complaint verbatim Add dates, names, and contact info (if possible) 60
Academic Payoff from Proof-of-Use Deep understanding of technical, operational, and economic aspects of the problem domain Your nuggets will unify into a sophisticated understanding of the domain You can now create value intentionally You can now find nuggets intentionally Provide improved products to users in the field, and leave them behind