Hi, good afternoon, good morning everybody. My name Manuel Garza, I'm a Director for the CTPAT program within the Office of Field Operations with US Customs and Border Protection.
Ari?

My name is Ari Schuler, I'm the Director of the CBP Innovation Team and advisor to Commissioner McAleenan. 
Stephanie? 

My name is Stephanie Smith. I serve as a scientific and technical advisor to the United States Postal Inspection Service.

And I'd like to introduce the Office of National Drug Control Policy, who couldn't be with us here today but are excited partners and looking forward to this challenge. Thank you. Luminary?

PRESENTER: Great, our agenda for today includes an overview of the challenge stages, timeline, and prizes; insights into the challenges of detecting drugs in the international mail environment; the specific aspects and problems that this challenge seeks to tackle; the selection criteria and submission process; and finally we'll end with a Q&A session. Note that this presentation should serve as an informational resource, and not as an authoritative document on the challenge. Please refer to the Rules, Terms, and Conditions for full challenge details. And now we will provide an overview of the challenge. 

So why are we holding this challenge? As many of us know, we're facing an unprecedented public health crisis. More than 100 people die from opioid overdoses each day in the United States. National Safety Council's analysis of preventable deaths identified opioid overdoses as the number one cause of unintended death. For the first time in America's history, Americans are more likely to die from an opioid overdose than a vehicle crash. International mail has been identified as a route for illicit opioids entering the United States. And is commonly transported in nearly pure powdered form, which leads to large scale drug trafficking via small packages sent in the mail. 

As a part of a comprehensive effort to address the opioid crisis, the United States government is seeking new tools and technologies to detect opioids in parcels moving through international mail. The challenge calls on innovators like you, in a wide range of fields, to help disrupt the flow of opioids into the United States. The challenge seeks well-developed plans for tools and technologies for quick and accurate detection of opioids in parcels without disrupting the flow of mail. Ultimately, the government sponsors of the Opioid Detection Challenge hope to work with challenge participants to commercialize their solutions and deploy them at scale in the field. 

Stage One launched on February 27th, and submissions will close on April 24th at 4:59pm. As soon as Stage One submissions close, finalist selection process begins, which we will cover with more detail later in the webinar. Finalists are expected to be announced in June. Stage Two finalists will participate in the prototyping accelerator, which will provide access to mentors, guidance from government experts, as well as educational webinars and other resources. During this stage, finalists will refine their plans and functional prototypes as they compete for the Stage Two prizes. Between Stage One and Two, finalists will have the opportunity to modify their teams. Stage One entrants will have an option to share their contact information with the Stage One finalists as part of the submission form. At the close of Stage One, we encourage finalists to augment their teams as needed to ensure they have the expertise needed for the prototyping accelerator. Stage Two will culminate in a live test event, in which finalists will demonstrate their prototype for the judges in a controlled environment at a DHS selected facility. The judges will evaluate the Stage Two entries based on their submission forms and tested on performance. These dates are subject to change, so please continue to check for updates on the challenge website, and subscribe to the newsletter to receive direct email updates. 

Challenge finalists and winners will receive cash prizes. The challenge has a total prize purse of 1.55 million US dollars. The Stage One finalists will each receive an equal share of the $800,000 Stage One prize pool. After Stage Two, the judges will declare a grand prize winner who will win $500,000, and a runner-up who will win $250,000. Additional benefits to entering the challenge include, publicity for finalists and winners, support from the government sponsors involved, and additional access to resources. Note that Stage One prizes are intended to be used as seed money for finalists to develop prototypes in Stage Two, and by accepting the Stage One prize, finalists are agreeing to participate in Stage Two. With this challenge overview in mind, Manny Garza and Rosanna Robertson will provide a deeper look at the nature of the detection challenge we are working to address. Manny will begin with a look at the opioid crisis.

Alright, thank you very much, Luminary. 

So good afternoon again and good morning to those of you on the west coast. You know, what are we doing this for? A lot of people have already been asking me why are we focused on fentanyl in the mail environment. I'll start off with the fact that every single day, CBP is processing in excess of 1.4 million small shipments every single day. I don't have enough officers in the field to look at 1.4 million shipments. I don't have enough canines to look at 1.4 million shipments. So we're looking for technology, we're looking for something new and innovative that can help CBP with this issue. As we keep moving forward, there's got to be a better way 'cause fentanyl, we're finding it as small as less than one gram in some packages, and we're finding it excess of one kilo in others. Keep that in mind as you're thinking about it. We've identified on the slides and online on the website, the specifics of what we're trying to get out of this, but know that there is a large variance between less than a gram and up to over a kilo of what we're looking for. And again most of it is in powder format. We are starting to see more and more pills that are being developed and identified as oxycodone or OxyContin, but they really are just fentanyl with some type of powder. 

As we move forward, we're looking at the avenues in how mail gets to the United States. So we've got opioid producers coming from overseas coming to the US via the mail system through foreign postal operators into our international service centers. I'd like everybody to be aware of the different tools that we're currently using at our international mail facilities. Upon arrival at our facilities here in the United States, all of our mail goes through radiation detection equipment — whether it's at the bag level or at the package level, and in some cases it's at the container level — of radiation detection. But for every instance of mail, we do have a focus point where everything can be inspected for radiation. As we move forward through our international service centers, CBP utilizes canines, we have x-ray detection equipment, and we have physical inspections of packages that we are conducting of those high risk packages. You should be aware that we are also looking at new technology such as CT systems that we may be installing at some of our mail facilities. We currently don't have any but would like for you all to think about that as we're moving forward. 

We have five main international service centers throughout the United States where we'd like to implement these capabilities at. Keep in mind that even though it's only five locations, we do have nine mail facilities. We have another 28 express facilities throughout the US, and the postal service has thousands of mail facilities throughout the US where this technology could be used at. As you're thinking about this technology that you'd like to design or provide a solution for in the inspection of mail, the most important thing that we want to accomplish out of this is that the package cannot be opened for the detection of fentanyl, opioids inside of the box. It's very important that it be in a fast process. Something that's quick as we want to be able to segregate these packages as fast as we can. Again 1.4 million, 80%, maybe not 80 but 60,% of all the mail is going through the JFK facility, so I really need to focus the effort on being able to find the high risk shipments. Whether it's at the truck or the container level, whether it's at the bag level, or whether it's at the piece level, we're looking for a solution that can help us with this. And the last thing I'm going to say to all you potential solvers out there, is that while today we're focused on the opioid epidemic, I'd like for you all to keep in mind that it should be a system or technology or something that's adaptable to what comes up next year, because five years ago the problem we had was with meth, today it's opioids, and in three more years we might be facing another battle. So please keep in mind what we're trying to figure out that can work today but can also be adaptable for the future. With that, I'm going to turn it over to Dr. Robertson.

Thanks Manny. 

So now that Manny has described mail operations, I'd like to better describe the types of solutions that we're seeking. As you learned, CBP receives and inspects an enormous volume of mail each day, therefore, we want you to focus on the mail which most frequently contains fentanyl, and that can be inspected and processed with an automated system, such as with the conveyance system as shown on the slide. The ability to inspect other types of mail will be viewed favorably by the judges, but it is not required for this challenge. 

Because of its potency, fentanyl and its analogs present the greatest danger as well as the greatest challenge for enforcement perspective. You heard Manny mention that as small as one gram or less can be seen in the mail, so we need a technology that can focus on small quantities as well as large. This slide demonstrates the potential foreign factors that creative, industrious criminal organizations and smugglers use to conceal narcotics. It's important that you consider that it's not just pills or powders, but that it can also come in other form factors that might influence how you detect it. As Manny just said, because of the large and growing number of fentanyl analogs, as well as CBP's overall mission, we seek solutions that have the flexibility to adapt and expand as the nature of the threat evolves. 

For your solutions, we are looking for any information that is gleaned from the physical parcel itself at the time of inspection, such as height, weight, appearance, handwriting, or even x-ray images are provided. Examples of these types of data are provided on the challenge website. Solvers who advance to Stage Two will be provided access to a more robust set of data, however, it's important to note that solvers may not rely on external data that is collected prior to parcel inspection such as information about the sender or recipient, or data collected from the government or private commercial entities and data bases. 

Lastly, for a solution to eligible for the prize, it must satisfy these specific requirements around function and size: pretreating and penetration of parcels is not permitted, you cannot open the package as Manny said. Solutions cannot exceed a size of 180 inches by 81 inches by 78 inches in size. The size of the parcel itself, or the size that you must be able to scan is 27 by 17 by 17 inches. Though smaller is okay. For a complete list of eligibility please see the rules, terms and conditions on the challenge website. 

Back to you Luminary Labs.

Thank you. Next we will discuss the process for submitting to the challenge, what we expect from solvers at each stage, how submissions will be evaluated, eligibility and intellectual property. First, we want to underscore that this is a global challenge and we encourage entries from anyone who meets the eligibility requirements. Our goal in Stage One is to identify well developed plans for solutions. These should detail the functionality of a solution and a path to creating a prototype. Additional details on what is required for submissions can be found on the submission form. Submissions will be evaluated by the judging panel according to the Stage One selection criteria, and up to eight finalists will be selected for participation in Stage Two and awarded cash prizes. 

The six Stage One criteria consist of the following: first, accuracy, or the ability to reliably find small amounts of opioids in parcels with limited errors. We're also looking for flexibility, which refers to how well solutions will adapt to change in the dynamic drug detection arena and how easily a solution could be adapted and redeployed accordingly. We're also looking for speed, which is in other words that we're looking for solutions that can keep up with the high velocity of parcel processing in the international mail environment, as well as usability, which refers to the ease with which operators can learn to use the solution. We're also looking at feasibility. We want to know the extent to which the proposed solution uses technically sound methods backed by credible supporting evidence and can be developed into a testable prototype within a four month accelerator period. And finally we're interested in the team. Judges will evaluate your teams experience and ability to move from plan to prototype should you be selected as a finalist. Each criteria will be given equal weighting in the scoring. 

In Stage Two, finalists will participate in a prototyping accelerator. Participants will develop their plans into prototypes and prepare for a live test event in which solvers prototypes will be tested before a judging panel. At the end of Stage Two, judges will expect to see functional solutions that can detect opioids in a controlled setting. The submission forms will also require further details regarding steps towards designing and implementing the proposed solution. Further information regarding Stage Two will be posted on the challenge website and shared by the challenge newsletter ahead of Stage Two launch. 

Our judges panel represents a broad array of expertise and deep experience in detection, forensic science, drug interdiction, postal operations, and artificial intelligence. Their backgrounds also represent extensive experience in product procurement, development, and implementation. Judges will review and rate submissions according to challenge criteria. 

The challenge is seeking solutions from solvers across the globe, and there are several exemptions to this, including organizations and individuals on the GSA excluded parties list, or on the US Treasury Office of Foreign Assets control list. Both of which are ineligible for participation. Felons and teams with team leads under the age of 18 are also ineligible. And finally, government employees applying as individuals or on behalf of authorized eligible organizations must first ensure that their participation is permitted. For the complete eligibility requirements, please see the Rules, Terms and Conditions on the challenge website. 

The key highlight regarding intellectual property is that each solver team which submits to the challenge retains full ownership of the submission, but the government reserves the perpetual license to access and modify the submission. Further, by submitting, you grant the government the right to enter into negotiations for rights to operationalize your solutions. These are summaries of the policies and do not convey the full details, and we encourage you to familiarize yourself with the full policies in the rules, terms and conditions. 

As a reminder, the deadline to submit your solutions is 4:59pm Eastern Time, on Wednesday, April 24th. Again, please submit at least one hour before the deadline. Entrants can only submit once they've completed all fields of the submission form, including the submission checklist at the bottom of the form. We encourage you to register on Luminary Lightbox well in advance of the submission deadline so you can begin preparing your submission. 

Thank you for your attention so far, and at this point we would like to open up to questions. We'll answer as many as we can with the remainder of our time, and for those we are unable to address, we will post responses to as many of the questions on the challenge blog as a follow up to today's presentation. The slides will be available in the coming days, and the recording and answers to these questions will also be posted at a later date. Keep in mind that everything you say in the chat window will be visible to all participants, and we will take your questions one by one. 

The first question is, "We have two potential solutions, one that would serve the inspection mission and one that would require the package to be still for a few seconds, and another separate solution that would serve the higher mission. Should we, may we, submit both technologies? And if so, should we submit them both in a single submission or separately?"

Yeah.

We'll hand that question off to Rosanna Robertson.

ROSANNA: Okay, I just saw concurrence from my partners here and it looks like for these two separate missions, we would encourage two separate submissions so that we can evaluate each technology on the merits that it's trying to fulfill in the mail inspection process. So two submissions.

ARI: We would also call to your attention the specifications that we put into the posted challenge, because there are some throughput issues there and if you cannot meet those it may affect your ability to move on.

PRESENTER: Thank you. We have another question. "What does the training material for AI and machine learning solutions look like?" 

There are currently sample images, x-ray images posted on the additional resources page of the challenge website that you can refer to, and additional images will likely be posted in the next several days.

Yes.

PRESENTER: We also have a question, "Who will supply the data?" 

The Department of Homeland Security has been supplying the sample images. 

PRESENTER: Next question, "What power and networking capability is available?" 

We will refer that question to the DHS team as well.

MANNY: So, I'm assuming this is in terms of electrical, it's got to be suitable for here in the United States. So whatever electrical power is available. Standard 110, but 240, and if there's any special requirements then we would work on it case by case. We don't have any technical standards for electrical grid on this. And then the second part of the question...

PRESENTER: Great we have... I'm sorry. 

MANNY: The second part of the question was on networking. I'm not sure what that means. If they're trying to get at if we need to network multiple systems together, or do we put it on a database and then tie it to our machines. Maybe more clarification.

ROSANNA: Internet I think, Wi-Fi...

MANNY: Wi-Fi network, okay. So those, we will have those capabilities available as well.

PRESENTER: Great, we have an additional question: "When will the unit be eventually deployed in the field? You may have covered this already, but what quantities will be procured for deployment?" 

We'll refer that to the government team as well.

MANNY: So this is Manny. I think it's really going to depend on the type of technology that's developed. Clearly if it's a piece of equipment that's going to have to be integrated into an automated parcel system, how fast does it work? How many automated systems do we need to include with it? There's a variety of questions. If it's a small device that can just attach to something, we can put them throughout multiple layers. There could be a need for more of these devices. Don't just think of this as a CBP, United States only problem. Our friends to the north in Canada, who I've been working with very tirelessly on this effort, are very eager to help and be right along with us on this problem and try to solve it. So there's a need just from our friends to the north up in Canada, but in Europe there's a wider need for something throughout not just customs, not just the postal service, but Bureau of Prisons is always looking for something in terms of identifying narcotics before it gets into the jail streams. All the state of local jails as well. This has bigger implications than just CBP. CBP handles a large portion of this, but don't just think of it for CBP only.

ARI: In addition, we are ensuring that we'll have the contract mechanisms ready and in place. So if they're particularly promising solutions that are ready to move quickly, the government will be ready to move quickly as well.

PRESENTER: Next question: "How will postal inspection services be involved with detection in the field?" 

We'll refer that to Stephanie.

STEPHANIE: Typically the ability to open parcels is more limited for the postal inspection service because of the sanctity of the seal and an expectation of privacy. This is why we're working so closely with Customs and Border Patrol, because international mail, until such time as it has been handed over by CBP, their capabilities are greater in terms of opening parcels without a warrant.

PRESENTER: Next question: "Are there characteristics on the outside of the mail piece by which risk packages can be identified? How do inspectors currently identify them?" 

We'll refer this to Manny.

MANNY: Okay, so characteristics. As we mentioned earlier, we don't want solvers to look at the name on the package, look at the address on the package, or where it's going, who it's coming from or what the description is. Those items CBP already has a robust system capable of doing that. Some opportunities I think that would be good for the characteristics of the package themselves is looking at a package, the shape, the dimension of it, the labels, looking at specific labels on a package, looking at postmarks on the package itself. If a system is able to look at a postmark of where it's coming from. What we would be able to do with information like is, as we find drugs inside of a package, how can we go into a system to do that? How can we plug it in that says, this package had drugs in it and here's all the characteristics. Then you go back and you look at that postmark, as an example, we're finding more and more bad guys sending a lot of packages that are very similar. So that could be very helpful as well.

ARI: And for Phase One, it's important to keep in mind that there are characteristics like Manny just described which can be indicative of illicit behavior, and in Phase Two we will be providing a data set which will show you what the true positives are, and allow for solvers to dive deeper in potentially looking at what truly identifies a true positive.

PRESENTER: "Will image data be of actual fentanyl samples or of surrogates?"

MANNY: So we actually have both. The images that are currently posted on the website are of actual fentanyl seizures as well as some meth seizures. I think there's one or two that are cocaine, but those are all narcotics. That's just to give you the representation of what it looks like for narcotics. But as we move forward to the Phase Two, there will be a lot more specific information that we will provide to those that make it past the Stage One.

PRESENTER: Great, next question. "Is structured data available combined with the image scan? For example, the height, the weight, etc."

MANNY: We don't have all the physical characteristics for specific packages. That may be something that we see if we can do during Phase One time period where we can take some samples and get photos and weights and dimensions, but it's not collected today.

PRESENTER: Next question is, "For packages in the mail streams, what is the size and weight range? What minimum size, what average size, what maximum size?" 

We'll refer that to Stephanie.

STEPHANIE: Well the good answer to this is there's tremendous variety. In terms of what we refer to as e-packets, they're very small, typically the size, probably the size of a five by seven flat, padded style envelope, but they could be as large as typical if it fits it ships parcels. I would keep in mind that we have provided limits for you in terms of the size of the parcels that we would expect the technology to be able to handle. In terms of weight, it's largely related to the contents of the parcel. Again in the category of if it fits it ships, you could have something very dense in a package that weighs quite a bit, but keep in mind that if the contents are something like fentanyl, it's not going to be that heavy. You could figure out from the density of a powder what the typical weight limits would be on a parcel.

PRESENTER: Next question is, "Is there interest in differentiating fentanyls from pharmaceuticals?" 

We'll refer that to the government team.

STEPHANIE: To be honest, even pharmaceutical products are limited to be shipped in international mail, so for our intents and purposes it doesn't matter if it's illicit or it's quote pharmaceutical. It's not supposed to be shipped through international mail so we want to detect it either way.

MANNY: Yeah and then on the CBP side, pharmaceuticals in the mail or just pharmaceuticals in general in any environment, if they're for personal consumption, that's really not allowed either unless there's a doctor's prescription attached with it and it would have to be the original doctor's prescription. So the likelihood of that happening is very slim. But I think at the end of the day we want something that can definitely tell us, this package is a high risk, or a higher risk than the package that doesn't have pharmaceuticals or opioids inside of it.

PRESENTER: Great, we have a couple of questions left and we will be watching the chat window. Please enter any additional questions you have and we will wait a couple of minutes as well to see if any additional questions are asked. The next question is, "How will we know if an international item has been processed through detections systems or just simply presume to have been? We see items now that have made it to the delivery unit only to be unclaimed by the recipient, and later determined to have contraband." 

We'll refer that to the government team.

MANNY: I'd have to do some more research on the question to understand it and we'll get back.

STEPHANIE: I can tell you domestically in the postal service if we do take an action to interrogate a parcel we have a process by way of indicating that it has gone through some level of screening. This is more in the area of our dangerous mail investigations program. I'm not sure that I completely understand why it would matter to an innovator about the ability to know whether or not something had been screened, and really if attaching some sort of indicator to the parcel was something that ultimately slowed down the entire process. We'd have to weigh whether or not the value added was great enough to mark those parcels in some way to indicate that they had been through the screening process. Which, just on gut I would say probably not that valuable.

MANNY" And I would agree Steph. I mean if we're talking about putting an indicator on the package just to let everybody know it's gone through this, the answer is no. That's not something that CBP is interested at this time.

PRESENTER: Next question, "Are there packaging materials that need to be factored beyond paper, cardboard, plastic, composite films and sheets? Is metallic packaging included?"

ROSANNA: Is mylar considered metallic?

STEPHANIE: Well, I think that the innovators should consider that if you're in the habit of sending things that you don't want to be detected through the mail stream, you will use a variety of techniques in order to hide the material from detection. Certainly I think it's safe to say that we have seen a tremendous variation in the way that people contain these materials, including sometimes forming them into another material to make it seem as though it's something that it is not physically, another material. I think you should consider as innovators that there are a lot of clever methods that people use, and another thing I would say is to the point that we've already discussed, being flexible, is that we certainly find that as criminals are caught in one way, they are amazingly adept at developing a new way. So we want to be able to be flexible to detect it in a variety of different packaging situations.

PRESENTER: Next question is, "For machine learning computer vision solutions, will there be examples of non-seizures to develop a classification model against? If so, how many?" 

There are currently sample x-ray images available on the additional resources page of the challenge website. We'll refer this question now to Manny for any additional information.

MANNY: I'm actually going to turn it over to Ari.

ARI: So as we've mentioned before, in Phase Two you will receive a sample set probably in the 1000 to 2000 image range with some associated meta data. Once you get to that phase there will be an NDA you have to sign because of the sensitivity of the data, but we believe it will be sufficient to advance your prototype through the Phase Two accelerator.

PRESENTER: We see a question, "Does this solution need to be image based?" 

The answer is no, it does not need to be image based. 

PRESENTER: We have a question, "Any cost ceilings at this time for the portable units that can be utilized in the field?"

MANNY: Can you repeat the question, I'm sorry?

PRESENTER: The question is, "Are there any cost ceilings at this time for the portable units that can be used in the field?"

ROSANNA: At this time no. We're considering all solutions. I think there is a requirement that mentions cost relative to annual operations and maintenance, but all solutions are considered at this point.

PRESENTER: There's a question of, "Whether Phase Two data sets will be x-ray images, or images of the exterior?" 

Ari?

ARI: It will be both.

PRESENTER: There's a question whether we'll be publishing the questions and answers from the session. 

Yes, we will be publishing this to the challenge blog. 

PRESENTER: We see some typing, and so we'll wait to see what questions come up. Okay, there's one. "Would a portable pipe unit that can only scan about six inches into the parcel be considered?" 

The answer is yes, we'll consider all entries from eligible teams.

ROSANNA: I think it's worth noting that if there are follow up questions after this webinar, people can use the email address hello@opioiddetectionchallenge.com.

PRESENTER: There's an additional question. "How would we know if an idea has been previously considered or has a patent, or partial patent attached?" 

The IP considerations of existing patents will be the responsibility of the solver to identify in advance. 

PRESENTER: Next question, "Will you consider online solutions as well as offline? As a second line of defense?" 

We'll refer that to the government team.

ARI: Yeah we need clarification of what you mean versus online and offline.

ROSANNA: If you're talking online as in using data that's open source or openly available, whether that's social media or dark web, then those are not solutions that we're entertaining at this time. But feel free to provide additional clarification.

PRESENTER: Next question, "Are you planning to work with the origin countries for their cooperation in interdiction?" 

We'll refer that to CBP.

MANNY: So we actually are already working with many countries on interdiction. We are running many operations with multiple countries to try to detect and interdict narcotics prior to their departure from the originating soil.

PRESENTER: Next question, "Will commercially available solutions that need limited improvements be eligible?" 

It is the responsibility of solvers to obtain any permissions that utilize existing technologies that have patents or other limitations and to secure that as a part of their submission. Are there any additional comments from the government team?

ROSANNA: No.

MANNY: Well, the only thing I would add is that if they are going to do any commercially available solutions that need tweaking, that you identify that in your Phase One solution. That you have permission from the commercially available source to use their product to move forward.

PRESENTER: We have received a clarification on a question regarding online versus offline. Their clarification is that, "Online means while the mail piece is in the incoming postal stream. Offline means image analysis on an image database once the piece has left the sorter."

MANNY: I would say they just submit their solution proposals. We'll be willing to review them.

PRESENTER: Well thank you everyone for your time today. If you have additional question you can send them to hello@opioiddetectionchallenge.com.

ARI: And we'd like to thank everyone for their time and participation and great questions, and also the many government agencies supporting this challenge. We really hope whether you've worked with the government before or never have and you're drawn by this mission that you will participate as a solver in this challenge and help us combat the opioid epidemic in our community. Thank you and have a great day.

Thank you.
Thank you.