Passenger Self Screening Industry Day Breakout Sessions Notes.pdf

Industry Day: 11 am Breakout Sessions

• Integrated Algorithms

• Analysis and CONOPS

• On-Person Screening

• Accessible Property Screening

Integrated Algorithms

Douglas MacIvor (TSA) Jennifer Koons (OPO)

Brian Lewis (S&T Contractor, SME) Laura Lyons (S&T Contractor, Program Support)

Session 1: 15 minutes—11-11:15 am

• General Opening Comments o Basic introductions on algorithm development

• Discussion o To mature systems into larger integrated network systems, TSA must problem solve creatively. One effort currently underway is Sandia National Laboratories’ (“Sandia”) Open Threat Assessment Platform (OTAP) focusing on middleware that is located on transportation security equipment, interfacing with algorithms, in support of TSA’s Open System Architecture initiative.

o In the context of algorithms, TSA could successfully integrate CONOPS and screening systems by applying algorithms to provide information to passengers (rather than just applying algorithms to provide information to Transportation Security Officers (TSOs), which is typically done today).

o The main goal of the above-discussed is modularization.

• Questions o What will happen to infrastructure in airports? Not all 440 airports have available pipelines.

Everything is on the table. A limited number of prototypes may be tried, to assess return on investment (ROI), with follow-up steps based on that assessment.

There are engineering approaches that may be useful.

o Can TSA collect particular data of passengers/body scanners legally?

No—TSA cannot obtain body scans of people.

However, portions may be able to be collected legally (not fully reconstructed body scans) in the future based on discussions with privacy personnel, if it is worthwhile for development

Yes, baggage data can be collected, under specific conditions and where no specifying information exists.

TSA will take privacy concerns into account to engineer solutions.

o Has privacy by design been factored into the concept of collection of passenger data?

Yes. TSA has a no-fly list, selectee list, PreCheck, in which people enter date of birth (DOB) and gender. This can be categorized into four risk-stratified groups of people.

There is a voluntary program with well-defined background checks (PreCheck) risk-informed screening that takes place today.

The Innovation Task Force has designed controlled scenarios on this effort.

o Has TSA considered synthetic data to solve data privacy challenges?

Yes. Screening at Speed (SaS) had synthetic data as an SBIR topic in FY19 and TSA is interested in the results from that effort.

o Does privacy data collected through TSA remain on the cloud?

Not exactly. There is a Chief Information Officer (CIO) in charge of this data.

The cloud is a future state item. Data is being tightly controlled.

o Regarding Sandia, is TSA building its (TSA’s) detection models in-house or not?

No. TSA does not have the expertise in-house.

There has been a past Prize Competition (DHS S&T and TSA’s Passenger

Screening Algorithm Challenge) which delivered eight algorithms for further development.

Session 2: 15 minutes—11:15-11:30 am o From an algorithm perspective, TSA is really interested in interfaces and how all industrial partners can collaborate to provide an interesting, productive experience to the passenger. The output goal is providing sufficiently-detailed information to a passenger.

• Discussion o Potential problems for people developing algorithms may include integration/who will do the integration, providing an interface, and so on.

o This complete effort would be challenging for one single entity—this will likely involve collaboration instead.

• Questions o What is the correlation between an X-ray system algorithm versus bag scanning?

TSA may want an X-ray algorithm to determine specifically when to scan an item during the overall scanning procedural process.

Are TSA’s algorithms further classified into separate algorithm systems now?

• No, technology has not been developed to this point presently.

o TSA needs somebody to own the overarching system, controlling how all of these interfaces are happening.

True, but there is a caveat from the RFI perspective. There could be a 3rd party integrator, or maybe also the X-ray system (or other component) happens to be the defining component. With so many possibilities, there is no good answer yet.

This is for the RFI—if vendors have ideas, tell TSA through the RFI.

Ultimately, TSA needs sensors that are both sensitive and specific.

o TSA is looking for an elegant solution—this is not hard—but the challenging aspects may be resolution and integration. Regarding the RFI, if TSA examines a PreCheck traveler—how many alarms data on the body and/or bag can be collected? Data informs many solutions.

There are a number of issues described in this question. Focusing on PreCheck initially is what TSA is doing. TSA has not published alarm rates for accessible property re: PreCheck yet.

There is a distinction between primary detection versus resolution. TSA has initial filters (X-ray machines) that are not optimal by themselves. TSA may need to meld resolution together in a seamless system (e.g., have a better algorithm as one method potentially). This depends on sensor quality and capability.

Differentiating between explosives and other items is challenging for existing sensors. TSA needs sensors to effectively pull/detect items out of the “noise.”

o How does TSA use the algorithms to address the sequence of operations? (e.g., clearing person then bag, or same times, etc.)

Send TSA ideas. Yes, how these pieces come together and who owns pieces of these algorithms is key to consider.

Session 3: 15 minutes—11:30-11:45 am

• General/Brief Opening Comments o Brief basic introductions on algorithm development

• Questions o What do integrated algorithms mean?

Integrated algorithms are essentially a complex system, more intricate than any transportation equipment ever acquired. Integrating a baggage scanner, human interface, and more is involved. Also applied are datasets, cybersecurity of system considerations, and network patching.

TSA hopes analytics will be a function evolving over time to start with an initial development phase, test items, and so on.

o System integration is challenging; instead of siloed algorithms, can algorithms have an order of operations?

In the current state, TSA has designed TSO-facing algorithms (which are by nature repetitive).

o There are many additional questions to consider, such as:

How does TSA use outputs of algorithms?

How does an algorithm give accurate information to passenger?

How does TSA factor in confidence to give good information back to the passenger?

The business problem of structuring incentives and examining the marketplace for vendors is crucial to consider; how does everyone collaborate for a complicated system such as this?

What role does TSA take? Is there one vendor who does the X-ray system but also owns system integration? These are questions that TSA wants vendors’ answers on.

o This integration model will remove certain aspects, but what about PreCheck/similar?

Is PreCheck meant to remove secondary screenings? Yes.

Longer answer: the better the algorithms and sensors are, TSA would not need secondary screening.

TSA may want to adapt to different cultures, mobility aspects, and there are many opportunities to get creative o Can’t TSA extend PreCheck/similar to include cultural/mobility aspects—so the system “knows me” and what trusted traveler program I am in?

Maybe. But it is not an automatic “yes.” Public opinion on use of privacy information is important to the TSA privacy office—transparency is imperative—showing what TSA is doing and how TSA is using privacy information.

But at RFI level, TSA is open to opportunities to improve the passenger experience; TSA would like to hear ideas, and factors in whether such ideas are truly useful, assessing the ROI.

o Is this integration on cloud or on premise? What about the running of an algorithm?

The latter—local computer, on premise, not cloud.

If there is a huge ROI for turning to the cloud, vendors, put that in your RFI.

o Airports are at different level of infrastructure—has TSA considered that?

Yes, TSA has considered this aspect.

Session 4: 15 minutes—11:45-Noon o Basic introductions on algorithm development o Algorithm areas of interest include: accessible property, the X-ray, CONOPS (how information is passed between algorithms), plus analytics to ensure everything works smoothly. Integration of these components is key.

• Discussion o This is the most complex system that TSA has ever acquired. Previously, monolithic systems were acquired. TSA manages those on an individual basis (not smoothly).

Upgrading systems requires going back to each individual vendor to increase their ability to detect a dangerous chemical, for instance.

o It would be extraordinarily difficult for one sole vendor to develop the whole system on their own (human interface for the traveling public, data collection, other algorithmic development). This is a significant data challenge.

o In the past, algorithms have been TSO-facing. Now TSA wants to take the output of those algorithms and provide them to passengers for a positive passenger experience.

• Questions o When TSA says “integrated algorithms” and “modularity,” how will TSA specify that in detail and how does TSA tell if one (AI) is smarter than another?

That is a hard question. TSA will work closely with the Transportation Security

Laboratory (TSL). TSA is in close communication with TSL scientists (even as recently as earlier this week).

Creation of the initial prototype in the field, followed by iterations, is the general idea.

The process TSA has today was applied when TSA acquired recently a system that went through a cumbersome testing process (cost/duration). Finally, contract award and deployment were completed on this system, but there was an object in stream of commerce testing that was not considered, and now there are countless of false alarm testing results arising as a result. TSA wants to move to a paradigm where TSA accounts for this; TSA would run algorithms in the background before flipping the switch for operationality.

o Can TSA turn going through security into a game (e.g., with leaderboard)?

Perhaps! TSA wants the passenger experience to be pleasant, not arduous—and TSA is always open to ideas. What has TSA not thought about that TSA needs to think about?

There are many strategic questions up in the air. The business relationships will be a real challenge, in terms of automatic screening vendors currently integrating with other aspects.

It is not the government’s role to dictate business relationships in the private sector. But—we would like to foster such relationships.

TSA would like to see teams as an integrated approach.

o What strategies would industry most prefer (how best work together) to come up with a solution?

440 airports—TSA would recommend at least 3 integrators in charge of

West/regional areas, for example.

Analysis and CONOPS

William Hastings (TSA) Carolyn Lethert (OPO)

Jason Deich (S&T Contractor, SME) Alesandra Trilivas (S&T Contractor, Program Support)

Session 1: 15 minutes - 11-11:15 am o TSA is considering the overarching aspect of the passenger interacting with the system while ensuring the passenger is performing correctly as well o What will the system do?

The system will analyze property.

If there are any items on a person that are prohibited, an alarm will sound since the system will see/notice it. Once the alarm occurs, the passenger will divest it by prompting the passenger to remove item and put in divestment bin (safety drop-box).

That item will be scanned and then will be cleared.

o What does this help solve?

This is so a passenger does not have to worry about TSOs looking through their bag or touching for screening.

o How does one ensure the process works effectively?

Video analytics. This will ensure compliance to directions for each section of the on-person screening section.

o How does each system work together and in harmony?

Specific video analytic systems correlate results from each part and monitor each reaction and monitor compliance.

There will be no additional staff to operate each on-person screening section.

o What if the system is unsuccessful?

One main option will be to return to staff screening.

• Questions o There was a comment regarding those with disabilities. What about parents with small children?

Currently, this system is not equipped for those with disabilities.

Right now, TSA is interested is getting feedback from vendors and what the challengers are for the passengers.

To obtain an initial look into how this would work, ITF is focusing on PreCheck passengers. Most PreCheck passengers are travel-savvy and single-adult fliers. TSA is interested in screening multiple people but not at this moment and TSA is unsure how feasible that really is.

There are security thresholds to meet.

o How does TSA currently do video analytics?

Video analytics has been evolving everywhere within DHS.

An example of the “dream” would be like what Amazon uses at their flagship store. Their flagship store has no checkout stations. The customer scans their Amazon prime membership and it charges them automatically from what they grab from the shelf.

This is a method where all parts are linked and that they work in concert.

o Is TSA looking into the commercial sector for customer engagement?

TSA is interested in applying human-centric design to this.

Some items they are considering include the overall system design, intuitive technology, the amount of physical space a passenger needs.

Goal is to ask these questions but currently, no technical topic category.

o Can a reply to the RFI include a technical topic category?

If one feels there should be more topic areas or be better defined or too restrictive, provide that feedback in the RFI.

o Has TSA considered medical devices that may not react well?

Just like with the disability inquiry, TSA does want to ensure the system fits many passengers.

Want to ensure that whatever solutions are applied can be applied to as many broad audiences as possible.

o Does TSA want to tie any accessible property to checked baggage to gain complete picture?

If an RFI submittal includes some suggestions with checked baggage in mind, a vendor should submit that still as that could be a part of a long-term solution.

Checked baggage is a desired capability and TSA is interested in including that with this idea; however, it may not be feasible at this time.

o There is a concern about privacy. What about wiping the AI information out of the systems at the checkpoint?

Biometrics will be integrated, but the main consideration is what legal restrictions come along with this.

Yes, biometrics will be integrated but TSA is uncertain of how it will be integrated and utilized as this is based off legislation and legalities.

o Is there a size requirement?

The standard checkpoint configuration is 54 feet long x 26 feet wide.

Each unit has no firm requirement. The smaller a vendor can make it, the better.

However, adhering to ADA compliance would be good, at 5 x 7 feet.

Doors would have to be 36 inches wide.

Not a hard restriction but if wondering about designing, vendors should start with those restrictions. These are already existing restrictions for TSA.

Session 2: 15 minutes – 11:15-11:30 am o CONOPS CONOPS need to be handled correctly by passengers.

• To have a door that opens on one side, the passenger engages with the system; passenger divests items into system; screens person in real-time;

and then continue divestment if requested by the system.

If there is a disruption, the system would tell the user that something is still in the pocket. The passenger would then be prompted to fully divest by putting it in screening system.

If the passenger can self-resolve, the system will prompt passenger to work on that.

Potential Solution to CONOP Fail: if the system itself is not working (for whatever reason) or if the passenger cannot complete each of the directives provided by the system, another door (or the initial door used for entry) would open. Then, the passenger would go through that door, and follow the designated path to another screening designated area. That next area would most likely still be considered primary screening but could include secondary screening as well.

o Analytics This is how TSA will ensure the passenger is following through on requests.

A video analytics system capable of correlating system and passenger is following directions.

Video will see the person removing the item, which will then prompt the imaging system to see that someone is not just moving the item from one area to another or throwing it out the door, so someone can pick it up afterwards.

• Questions o Why target PreCheck first, especially since the line is short? Should TSA move beyond PreCheck?

There is a strong interest in being able to screen any person.

Focusing on PreCheck because throughput is based on divestiture so PreCheck allows for that to go much more quickly.

There is a strong interest in having an overarching system that integrates multiple systems.

PreCheck passengers are well-suited to point technologies there for right now and initial testing concept of operations.

TSA wants to ensure that systems can be matured in the future.

o Is there a possibility of trace detection?

As of right now, no, as this is for the primary screening system.

If there are ways of doing trace or secondary screening, TSA is interested in hearing this and how it can be adapted to this environment.

If a trace solution that could be deployed could also resolve alarms quickly, that would be welcomed.

o Pod Concept

Is this example the best way to do this? This pod is all nice and pretty but what if a vendor has different design? TSA is okay with that.

When looking at overall system design, vendors need to remember that TSA is not getting any more space at checkpoints and not getting more staff.

TSA wants to be able to screen the same amount of people as they do now.

Train officers to be focused on alarm resolutions for workforce applications.

o Can TSA talk to the airports about power and weight?

This was a problem with AITs and CTs. Through work with those, TSA now knows tolerances on airports and knows upper-level thresholds.

How much can airports actually handle insofar as weight, size, etc. of systems?

• This is in reference to the AIT systems and the issues into which airports ran when TSA initially rolled out those systems.

• To potentially address this question, TSA has engaged with Dale Mason checkpoint architect.

o Why is TSA specifically focusing on checkpoint instead of a more dispersed solution?

S&T has a specific focus on dispensed sensing. This is the last stage before a passenger goes into the airport.

What’s technologically feasible? TSA is focused on checkpoint for immediate resolution.

o Comments about putting passengers “in a box” literally.

From a human factors perspective, how big to make the box comfortable?

Can TSA have systems that close and/or that far without issues and or with a great scanning ability? Individual feedback for each divestment and section.

Human factors: if someone does not want to use them, will not matter how great it is.

Session 3: 15 minutes – 11:30-11:45 am o There is an emphasis on not restricting oneself to the ideas provided at each table.

o Checkpoints are not getting any bigger and there is no more additional staff.

o Start engaging with the divestment system; allow it to complete the feedback and review.

o There was a discussion on the current divestiture process and how this possibility allows passengers to self-resolve as they can put property into property screening system and send to matrix system and review it in system.

o Analytics imaging could provide additional feedback to passenger.

o TSO is not needed unless there is danger with someone. If a passenger cannot self-resolve: CONOP may simply be to tell them to return to the screening process.

o Video analytics system is something that can observe a passenger and ensure they are acting in good-faith.

• Questions o Video analytics: Is there any benefit or thought on facial recognition? Compared against a watch list?

Biometric integration: considering identifying an individual and who they are

(known trusted traveler, etc.)?

Do we have biometrics on file already?

Will TSA also consider behavioral analysis?

These would be great information in an RFI.

o Biometrics integration (participant discussion) The biggest restriction is legislation.

Still interested if there is information on how to integrate biometrics into these systems.

An idea for a current prototype is for PreCheck population to use the on-screening option; confirm biometrics at entrance door; and it will not allow a passenger thru unless you are precheck passenger.

Interface and graphic user interface will be produced throughout.

Biometrics could also assist with language barriers. For example, if the passport used is from Germany, the machine would talk to the passenger in German.

o Is there data governance or data architecture that TSA will involve?

TSA does have some standardized data governance, but they do need someone in systems integration role.

If one person working on the project is a systems integrator, that means all associated parts should be able to be pieced together like pieces of a puzzle. This does not mean that the systems integrator would be responsible for the actual manufacturing of the system but rather how to put everything together.

o Let’s assume there is a video analysis system that can track movement; repetitive actions; would there be an anomaly detector if same movements are made for divesting, etc.?

Yes, the video analytics systems would have fool-proof methods to ensure that any anomalies with repetitive actions by passengers, etc. are noted. The system itself would produce an independent analysis of the actual action of divesting items, not just moving arms or hands to remove those items. However, if the system thinks something is unusual, the system will show the next prompt whether that is for a TSO to come to the system or for the passenger to exit and continue to another screening system.

o Discussion on envisioning small items going with bags Vision for this concept: reach up, open system, etc. open in particular way to open system; imagine tabletop surface and when you close the top, would put everything next to each other.

If vendors can make system perform intuitively, that would be great!

Session 4: 15 minutes – 11:45 AM-12:00 pm

• General Opening Comments o Video analysis can be used for observing a passenger and alerting TSOs if someone is not following the process. An officer only needs to be alarmed if the person is not listening.

• Questions o Fix to Alarm

If there is an alarm a passenger cannot resolve, what happens? If the system feels a TSO is needed, it will alert a TSO and keep the passenger in that space. TSO would be able to see what type of alarm happened and what type of threat may be present via the outside wall.

o Alarm Resolution Can this happen in same space and how? Include that in RFI.

TSA is moving forward based on what industry is telling TSA.

o What about so many items being in a passenger’s property?

From a CONOPS standpoint, the system may tell passenger to separate electronics and rescan property. In any case, the system will always tell the passenger what to do next.

o What is the potential layout of the system?

When TSA is discussing the x-ray system design, TSA is thinking of a cabinet style system.

Imagine tray looking thing and table top surface but dipped so it can accommodate that property; put property into space, it moves back for scanning;

it scans and gets the go-ahead.

o Miscellaneous Points PreCheck passengers: self-screening will occur more with this designation; the design and system will give no directions if passengers follow it all correctly.

This should be good for intuitive passenger use.

The system itself does not engage passengers unless absolutely necessary.

Regarding sizing: defer to ADA standard numbers.

The system is a very interesting concept as this idea is cost-avoidance and cost shifting. This will displace the cost of staffing.

System lifecycle cost: staffing is the most expensive cost. TSA wants an economically sound product.

There could be some contamination, which is a potential challenge.

Accessible Property Screening

Jason Hull (TSA) Duane Schatz (OPO)

Michelle Weinberger (S&T Contractor, SME) Chuck Meier (S&T Contractor, Program Support)

• Discussion & Questions o Does the government envision a prescreening resolution process to check divestment levels, or will the focus be on secondary resolution of screening?

The government has not decided yet; expect the system to give feedback if there is something that needs resolution. TSO interaction may occur after.

o Could the system give feedback by showing the raw screening image to passengers?

One could potentially use an advanced threat recognition (ATR) algorithm that shows the general threat location on the bag (like how body scanner images are displayed) and use audio and visual abilities to convey feedback.

o Is there a preference between making the system CT or X-ray based?

The government is not prescriptive in terms of technology; rather, the focus is on detection capabilities.

o What does the passenger do if the system detects an anomaly? (Rescan the item?

Call a TSO to investigate?)

If this program targets PreCheck (seasoned travelers)—if there is an anomaly, it could mean the traveler has to go through the regular security process, but that decision has not been made yet.

o What is the cost base for these systems?

It would marry several different screening technologies, so it will probably cost more. Full-time equivalent (FTE) savings could offset higher costs for each system.

• It is too early in the design process to get more detail on how much systems could cost, and what the FTE savings may be.

o What level of passenger training is the government interested in providing?

If the passenger is responsible for screening themselves, they may need some level of training to avoid common problems, but the government has not chosen a methodology.

• PreCheck passengers are the initial target audience, but this passenger population does not really need training.

• These passengers are already the fastest travelers; any resulting gains in throughput could be small initially.

o Is TSA going to take a position on open architecture?

There is increased interest in open architecture and standardization, particularly on the R&D side. Digital Imaging and Communications in Security (DICOS) is a good first step; increased standardization is of interest to increase integration.

The government is not prescribing a solution or an approach at this point.

• If TSA wants to mandate an open architecture, it may need to specify standards (Open Platform Software Library (OPSL, part of

DHS)/DICOS).

o What areas could cut the cost of the screening equipment?

Change the detection standards.

• This is unlikely to happen given the persistence of emerging threats.

Can we reduce the cost of hardware?

• Computed Tomography (CT) is expensive, but cuts cost by reducing actual searches on the back end, saving FTEs.

o If a passenger is PreCheck, what will happen differently from the general traveling public in terms of accessible property screening?

Initially, the program will be exclusively PreCheck, then will be flexed/adapted to different passenger types and situations.

o How large of a subset of the traveling public is pure PreCheck/Known Traveler

Number (KTN) holders?

The subset is 32-35 million total and growing approximately 200,000/month including trusted traveler and renewals.

This constitutes 10-20% of travelers per day.

Why do these numbers matter?

• The significance stems from the fact it seems as though the transition from pure PreCheck to broader audience will be the most difficult part of the program.

o How do we make the process intuitive?

The challenge will be to make the process simple enough for general and infrequent travelers.

How do we create the appropriate interface/feedback?

• The system needs to be audio and visual with options for multiple languages.

• Biometrics may help anticipate language and other settings or preferences.

o Or, the system can be tied to a pre-created profile that the PreCheck passenger creates.

• A TSA agent may be able to identify struggling and/or slow passengers and direct them to traditional lines or designate faster passengers as eligible for the self-screening.

o Would it be beneficial to look at the commercial sector to see how they have created easy user interfaces?

Is the concept that the systems have CT not Advance Technology (AT)?

• No, the focus is the capability to mitigate risks

• CT has certain advantages, but a specific technology is not the focus.

Rather, performance and capability is the focus.

o What are the challenges the government has not thought about yet?

The process could be simplified by sharing data across systems/sensors.

What people carry as accessible property changes rapidly (e.g., new cell phones/devices, changes in clothing every season). The system will need to be able to adapt to that.

o What are things that the government needs to take into consideration? Challenges?

Unasked questions?

Industry does not know what baseline the government is operating from, and what technology the government looked at or considered.

• The government has not prescribed a technology/methodology.

o How much latitude does industry have in developing a concept that adapts the old accessible property CONOPS?

Industry has broad latitude because there is not a prescribed technology or methodology. The focus is on the capability to detect threats, looking at APSS

6.2 Level 0 detection capability.

o What types of carry on threats will the program focus on?

Metals, powders, liquids, organics Currently using AT/CT to detect organics on precheck passengers o What is the appetite to allow passengers to be separated from their bags for a period? Could there be a possible bank teller tube system to scan bags?

The government is not prescribing anything yet; there is not a chain of custody policy that has been mandated.

o The TSA may need to involve other stakeholders (airlines etc.). How does the government stimulate and/or coordinate those relationships across the network?

TSA is responsible for relationship building across all aspects of industry in the transportation system; it is a government responsibility to coordinate these relationships.

o What sensors, technology, and/or architectural standardization coordination needs to happen to make this process work?

This is not actually self-screening, it is making the passenger comply with instructions and the system does the screening.

This is a business case; TSA should consider the broader industry to figure out the cost-benefit analysis in terms of increased costs per system and reduced FTEs over time.

o Is the idea to reduce FTE or to increase throughput?

Both, while maintaining detection capability.

o Is TSA looking for other ways to increase the PreCheck population, like giving passengers pre-check through means other than the current process?

Yes, the government is also looking to identify middle risk passengers between the traveling public and PreCheck.

Making the system intuitive will be the largest challenge.

• Discussion & Questions o The key challenge is to keep the detection levels while also adding this new concept.

The government is methodology and technology agnostic. The government cares about capabilities.

o Has there been any thought about sharing hardware and/or sensors between systems?

TSA has not yet defined any concepts; the government cares about any concept that meets detection standards and throughput. TSA has not closed the door on anything.

o If there is a threat that cannot be resolved in the bag, will the TSO come in, or will there be a secondary screening process within the pod?

The government has not made a decision about this. The system can provide feedback to help resolve alarms. How should the feedback be provided to help reduce the number of TSO interventions?

Giving multiple opportunities to resolve alarms may provide opportunities for hostile actors to probe the system.

There will probably be a need to focus on human education and acclimatization.

Not losing sight of bags will probably be very important to passengers; baggage should be made visible at all points in screening process. Doing this would foster increased passenger trust and reduced instances of lost items.

o Is the current CT capability suitable for this concept? How will the system react to false alarms? Just kick passengers to the regular screening checkpoint?

CT could be a solution, but the government is not prescribing any one technology or methodology to accomplish this task.

o What is the turnover percentage for TSOs?

There is a huge variance across airports. Smaller airports are often very low;

large airports can be over 100% per year.

o Should the government own scans and/or data of the bags to help track them across airports?

The government does have and keep scans of bags, with processing through

DICOS.

On-Person Screening

James Lambeth (TSA) John Whipple (OPO)

Nodari Ivanov (S&T Contractor, SME) Ryan Robison (S&T Contractor, Program Support) o What are other modalities out there that would be useful, (e.g., mmW), for the general population?

As vendors saw the form factors for right here, TSA has flat panels.

o What is the deficiency with Advanced Imaging Technology (AIT)—is it cost or speed?

Speed is the biggest issue—comfort as well. Most people would prefer multiple chances to divest in lieu of a pat-down. Ultimately, it is a matter of personal preference.

o TSA is starting this new system initiative with PreCheck, where the average is wait time is five minutes. How much throughput is TSA trying to gain by putting this at PreCheck? I thought the issue was with the standard screeners.

The larger issue is with standard screeners.

The current setup does not have traditional body scanners, allowing more freedom with where the system can be deployed—more freedom in the technologies TSA can use if something was at a lower TRL.

The idea is that when people see how easy it is to use the technology, they will be more inclined to sign up for PreCheck.

• This will, in turn, impact the cost factor by reducing the number of TSOs needed per lane—useful for smaller airports in particular.

o Would this method be paced at the same speed as walking through a metal detector?

The idea is that one of these pods would be slower, but more would fit in the space. Cumulatively, this would result in the line moving at the same speed or faster.

o Does TSA envision automatically closing doors or more of an open structure?

TSA is flexible as long as vendor-proposed ideas meet ADA safety requirements.

o With this, TSA is going to get passengers faster from the terminal to the seating areas, but passengers still have to wait for their carry-ons, right?

The system would allow passengers to bring their carry-ons with them while they divest. A potential solution is to have the bag simultaneously screened (via X-ray, CT, etc.) while the passenger is screened by a separate technology.

o Why does the property screening have to be part of this system? Why can’t TSA have a singular property screening system like in the current form factor TSA has today?

It all comes down to speed; the current slowdown is with accessible property.

The property screening aspect allows an individual to divest multiple times (i.e. if you have something in your pocket and it triggers the alarm, you can place it in your accessible property which will be with you). This will require rescreening of the individual. It is possible that accessible property (AP) will not screen until the passenger is cleared.

o In this mockup, there is a lot of confinement, it seems, with the closed door and whatnot; is that really a requirement?

No. There is no set of requirements; this is just an example.

TSA encourages vendors to look at different CONOPS, form factors and consider the impact on the following:

• Throughput

• Cost

• TSO requirements o Has there been any thought on how to avoid a person who is being cleared from coming in contact with someone who has not yet been cleared?

TSA would like that to be addressed in the RFI o Metal detection is a threshold capability; what opportunities could be explored to add more sensing modalities?

Thermal imaging comes to mind, which may be good for this instance due to the confinement aspect.

This is, however, not to say that we cannot use multiple technologies to cover gaps.

TSA is focusing on the PreCheck population because they are familiar with the process, and as such, allows more options in terms of detection-based technologies.

o At what point does the passenger get rid of their bags?

The idea is that the bags will come in the screening area with the passenger. Bags will be screened independently while the individual goes through their individual screening process.

o Does TSA think it is possible to use the body’s natural motion as a detection method?

The technology of mmW is already proven to detect a wide number of threats.

This may be a matter of having a form factor that does not require the passenger to stop. It is possible that panel-based form factors could fix this.

o What about individuals with metal in their body (i.e. for a medical reason)?

A metal detector is not a requirement, but obviously that would need to be considered. TSA would like to address as much of the traveling population as possible and does understand that there is a percentage of the population that would be unable to use this. TSA would employ traditional screening methods for these individuals.

o Does TSA envision when a passenger throws the bag in that it disappears and is cleared in another way or does it get returned immediately?

The idea is that accessible property is with the passenger at all times. This allows for passengers to divest on the spot if they have something on their person that alarms the system.

The underlying concept is that everything is self-contained. Nobody is comfortable with being separated from their accessible property; this is a key principle of the overall passenger experience.

o Is it a foregone conclusion that mmW technology is the primary mode of screening that would be used?

No, TSA is open to existing and emerging technologies. The draft SOW reflects

Technology Readiness Level (TRL) 5: not a constraint; if an emerging technology is at TRL 1 but can meet the timeline, that will be considered as well.

o Could this system hold multiple people such as a family?

The plan is for one person at a time. It is possible that small children go through conventional screening or TSA creates a system that would allow multiple people at a time.

• Additional Discussion o False alarm rates will likely drive technology.

o One TSO monitoring eight lanes cannot realistically be running all over the place if there are false alarms left and right.

This drives the idea that the passenger could divest themselves multiple times and handle the situation without TSO presence.

There would be no technological limitations for mmW to be continuously scanning.

With monitors and communication systems implemented, a TSO could remotely assist passengers.

• General/Brief Opening Comments

• Discussion & Questions o If it’s self-screening, do some of the privacy issues go away?

o What would be the limits of self-alarm resolution for on-person screening. Say a belt or a bra? How’s the TSO going to deal with it?

If the passenger cannot self-resolve, the passenger would step out and undergo the conventional pat-down.

o What if there was a robot conducting pat-downs?

If an alarm is sounded, the passenger would step out and undergo the conventional pat-down.

At what point do we require an individual to step out and divest?

• The general public is unlikely to respond positively to a robot pat-down, and this is likely to produce a negative passenger experience.

o In the concept provided, this is a closed space process. Has TSA thought about whether passengers will be comfortable in an enclosed environment?

That is where the human performance aspect comes into play. TSA wants to ensure consideration of federal regulations, passenger comfort, and the amount of space required.

Heathrow has an automated passport gate, and TSA could use similar ideas to create an open design. Looking at this, our question is can we get a walk-through metal detection system in there?

o With the pod design, would TSA have communications and remote monitoring so that a passenger could hit a “help” button and not have to step out for every issue?

This could be as simple as having an avatar and AI. The pod design could also have cameras within to monitor behavioral patterns.

o Does TSA believe that it is possible to use natural motion to screen?

The difficulty is not being able to control passengers’ movements. Introducing a universal motion such as opening a bin for accessible property could be utilized.

TSA has to consider the variance in passenger strength and capability (e.g., elderly and disabled individuals).

o Metal detection is a threshold capability; what opportunities could be explored to add more sensing modalities?

One option is a flat panel. Other ideas are anything that came out of the

Capability Analysis Report (CAR) that evaluated all future technologies currently in existence. This report looked at feasible technologies that would make this system solution viable.

• mmW, electromagnetic radiation, or flat panel may be appropriate solutions o One thing that I’ve experienced is being stopped and having my bag checked for explosives; a sniffer in the pod would eliminate a passenger alerting a secondary alarm once they’ve been initially cleared Yes, incorporating some secondary screening methodologies would assist with alarm resolution.

o Would TSA route passengers with families for items such as milk, insulin and/or other sensitive liquids?

Additional information could be gathered for, as an example, a breast-feeding mother. There would be assistance available to help steer the machine to avoid false alarms.

This raises privacy concerns; the machine could be something passengers opt into.

Language requirements have not been set; information could be gathered ahead of time. Generally, anything personalized improves the passenger experience.

o Is TSA’s main purpose in implementing this to eliminate TSO positions?

No, this initiative addresses smaller airports more than larger ones. A PreCheck lane at a small airport is not cost-effective. The primary driver is throughput and the overall passenger experience. The ratio of TSOs to passengers would decrease.

One thought to be leery of: TSA does not want self-screening to “give away” all the answers. The danger is that the system could increase predictability, making it easier to identify weaknesses for adversaries.