JA2018040277_FE_FBO.pdf

Justification and Approval FAR Part 6

JUSTIFICATION FOR OTHER THAN FULL AND OPEN COMPETITION

41 U.S.C. 3304(a)(1)

Pursuant to the requirements of the Competition in Contracting Act (CICA) as implemented by the Federal Acquisition Regulation (FAR) Subpart 6.3 and in accordance with the requirements of FAR 6.302-1, the justification for the use of the statutory authority under FAR Subpart 6.3 is justified by the following facts and rationale as follows:

1. Agency and Contracting Activity. The Transportation Security Administration (TSA) Contracting and Procurement (C&P), Requirements and Capabilities Analysis (RCA) and Acquisition Program Management (APM) propose to issue a contract without providing full and open competition. The C&P tracking number for this document is JA-2018-04-0277.

2. Nature and/or Description of the Action Being Approved. The purpose of this action is to provide a justification and obtain approval to award a continuation of an existing effort as a firm fixed price (FFP) contract for a period of twelve (12) months to expand development efforts of the vendor-neutral Automated Threat Recognition (ATR) software which was developed under the current contract HSTS04- 13-C-CT7000 with Stratovan Corporation, that was competitively awarded under BAA HSTS04-13-R- CT7000. This effort is considered the continuation of an existing effort of the current contract which provided the development and design of the ATR software. The effort is anticipated to cost $3,028,016.01. Both Program Offices (APM and RCA) are contributing funds for Phase III efforts. APM is providing $2,681,000.00 under PR#2118208AP1138 and funding in the amount of $347,016.01 is provided by RCA under PR#2118208ORC210.

Under the current contract, Stratovan was tasked to develop an ATR algorithm to meet the current detection standard utilizing Explosive Detection System (EDS or CT) images for a Smith Detection CTX- 9800. This effort was a proof of concept to determine whether a third party could develop an algorithm which would be vendor neutral. Under a previous contract, Stratovan developed a Software Development Kit (SDK) that allowed for vendor proprietary CT formatted data to be converted into a standard imaging format based on medical CT applications. Stratovan utilized the SDK to convert the EDS vendor proprietary CT format data into a standardized image format for Digital Imaging and Communication in Security (DICOS) and created the vendor neutral ATR, which successfully passed the detection evaluation.

The proposed contract will update the previously developed SDK to the new DICOS 2.0A standard. The adapted version of the SDK will provide the tools necessary for third-party development. The DICOS SDK will consist of a software library and conformance testing suite that facilitates the conversion between native security vendor data with the DICOS 2.0A standard.

The second effort is to expand the detection capability of the algorithm to include home-made explosives (HME) and integrate the algorithm on-board to the EDS machine. Once proven successful, TSA will be able to move into a competitive environment for 3rd party ATR development in the future, which will enable TSA to address emerging threats with more resilience and to gain innovative solutions existing in the open market from third party developers.

3. Description of Supplies/Services. TSA leverages technology advances where possible to meet aviation security challenges. Based on the latest Emerging Threats (ET) algorithm testing, TSA has established that the most effective way to continue enhancing algorithms while mitigating the risk of a false alarm is through machine learning technology. Stratovan is the contractor that developed and designed the current ATR software utilizing machine learning technology. Stratovan is currently the only source with the knowledge and capability to further develop the ATR algorithm to a vendor-neutral capability for EDS systems. Updating the SDK, Compliance Checker and Certifier to the most recent DICOS standard will provide TSA the tools to be able to implement the DICOS format which will ultimately become the foundation to allow third party algorithm development. TSA anticipates that, upon conclusion of this proposed contract, any follow-on support will allow for a full and open competitive process.

Performance Period – 12 months Type Total Price

Total Estimated Contract Value $3,028,016.01

4. Identification of Statutory Authority Permitting Other Than Full and Open Competition. The statutory authority permitting other than full and open competition is 41 U.S.C.3304(a)(1) implemented by the FAR Subpart 6.302-1 entitled “Only One Responsible Source and No Other Supplies or Services Will Satisfy Agency Requirements.”

5. Demonstration that the Nature of the Acquisition Require Use of the Authority Cited. Full and open competition is not a reasonable alternative for this requirement as it introduces significant risk to the Government. As stated above, Stratovan is the original developer of the current algorithm ATR and the only source with the immediate capability for a vendor-neutral solution. In the event the Government would have to revise its strategy and compete the requirement for the added development, the result would be that TSA will become reliant on current EDS manufacturers to develop algorithms. TSA has repeatedly contracted with existing EDS manufacturers to develop algorithms but these vendors have not delivered nor do they have a desire to propose a vendor-neutral machine learning algorithm.

6. Description of Efforts Made to Ensure that Offers Are Solicited from as Many Potential Sources as is Practicable. In 2013, TSA released a full and open Broad Agency Announcement (BAA) HSTS04- 13-R-CT7000 on Federal Business Opportunities (FBO) for a vendor-neutral machine-learning ATR algorithm and only one company, Stratovan Corporation, provided a concept paper in response to the BAA posting. Upon receiving the Government’s request, Stratovan submitted a full proposal which was evaluated as “acceptable” by the Government and resulted in a contract award (HSTS04-13-C-CT7000).

As Stratovan remains an expert in machine learning algorithm development specific to explosive detection, Stratovan is considered the only responsible source that can currently satisfy Agency requirements. Contract number HSTS04-13-C-CT7000 expires on September 30, 2018.

7. Determination by the Contracting Officer that the Anticipated Cost to the Government will be Fair and Reasonable. The Contracting Officer will determine the price of this effort as fair and reasonable based on the comparison to historical pricing with Stratovan Corporation under contract HSTS04-13-C-CT7000 as well as a comparison of the Independent Government Cost Estimate (IGCE) and the United States Bureau of Labor Statistics (www.bls.gov) website to compare the labor escalation rates for service-type requirements.

8. Description of Market Research. Market Research is based on the aforementioned BAA for a vendor-neutral Automated Threat Recognition (ATR) algorithm solution employing machine learning technology, as well as an assessment of the current algorithm capabilities of vendors supporting the EDS Qualified Products List (QPL). Submissions that were evaluated as “acceptable” for the BAA and eligible for award were required to demonstrate an expertise in conversion and employment of DICOS EDS file formats, EDS algorithm development and employment, and machine learning capabilities.

The Electronic Baggage Screening Program (EBSP) has manufacturers that are qualified to produce EDS machines and their respective ATR algorithms, such as Smiths Detection, L-3 Communications, and Reveal (Leidos). However, their current ATR algorithms are designated as proprietary intellectual property and would require TSA to obtain new EDS machines and algorithms via limited-source procurements from these three vendors. TSA encouraged vendors to use third-party algorithms, but vendors have developed their own in-house algorithms which are protected through intellectual property rights.

As a result of internet searches, vendor engagement, and knowledge of the industry by the program office, it is known that Google, IBM and Amazon all have the capability for machine learning. However, they do not have the capability for the specific type of machine learning required to develop a detection algorithm to screen checked baggage. Additionally, Google, IBM and Amazon have not developed any effort on the DICOS platform. Stratovan is the only source that has developed an algorithm utilizing the DICOS format. DICOS 2A is a TSA adaptation of DICOS 2.0.

To further illustrate the difference between the capabilities of the aforementioned companies and the capability of Stratovan and the DICOS platform, there are two high level categories where deep learning algorithms can be categorized; algorithms that detect objects (e.g. identification), and algorithms that detect material discrimination. The algorithms that focus on object detection are tailored to look at shapes and the form of objects, while algorithms that focus on material discrimination ‘see’ material properties rather than shapes and objects. In the checked baggage environment, it is crucial that the algorithm has the capability to ‘see’ material properties instead of objects. This is imperative due to the reason that explosives can take any shape or form and it is very difficult to enable an algorithm to distinguish between a common object and an explosive without the ability to ‘see’ material properties.

Under the current contract, Stratovan has successfully developed an algorithm using the deep learning methodology. This algorithm utilizes the deep learning methodology to enable material discrimination which is crucial for the detection of explosives in the checked baggage environment. Material discrimination is essential for the checked baggage operations environment.

9. Any Other Facts Supporting the Use of Other Than Full and Open Competition. Other third-party commercial providers do claim to possess machine learning and deep learning capabilities. However, the companies do not possess ATR development expertise nor the required EDS backgrounds to develop an ATR algorithm capable for checked baggage operations. The ATR development for checked baggage requires unique expertise in explosives, TSA requirements, data collection methodology, scoring, and EDS operation, specifically Computed Tomography (CT). At present, the only vendor that has the capability for deep learning ATR algorithm development for checked baggage operations is Stratovan who has already successfully built a vendor-neutral ATR algorithm using the DICOS platform. While it may be possible to invest and develop an EDS expertise with a new company, to do so would entail a significant duplication of cost to the Government and a substantial set back in time leaving TSA without the ability to have a third party vendor neutral algorithm for at least three years. By awarding this contract, TSA will be in a better position to enable a competitive environment for third party machine learning algorithm development. Although the Government owns all of the data rights to the developed ATR algorithm and current SDK, it cannot be used by another company for the current effort without causing significant delay, added expense and risk. This contract will enable TSA to level the playing field with respect to all screening equipment and will eliminate the proprietary nature of the EDS systems which limits interoperability.

10. A Listing of the Sources, if Any That Expressed, in Writing, an Interest in the Acquisition.

As previously stated, Stratovan was the only company to submit a concept paper under the BAA. They maintain expertise in machine learning algorithm development specific to explosive detection. Stratovan is considered the only responsible source that can satisfy Agency requirements.

11. A Statement of the Actions, if Any, the Agency May Take to Remove or Overcome Any Barriers to Competition Before Any Subsequent Acquisition for Supplies or Services Required.

The award of this contract will update the previously developed SDK to the new DICOS 2A standard thus providing the tools necessary for third-party development, complete the development needed to expand the vendor-neutral ATR algorithm to include HMEs and integrate the algorithm onto the EDS machine. It is TSA’s intent to solicit third party algorithm development in a competitive environment. As previously stated, the DICOS toolkit allows both new and existing vendors to become DICOS compliant and enables them to introduce new products into the airport security environment more quickly.