IT'S TIME FOR AN AUTOMATED DETECTION ROADMAP

The coronavirus pandemic situation is dire, and we need rapid problem-solving (not to be confused with the solving of all problems.  Yes, I’ve worked with many start-ups).

The coronavirus’s silver lining is that it heightens the urgency for fully automated detection at airport checkpoints around the world, paving the way for self-service security (first developed by the Dutch, the aviation security rock stars that they are).

Let’s break this down.

Efforts to automate detection for cabin baggage screening fall into two categories:

• Automated detection of explosives. Technology manufacturers have been working with government partners for close to two decades to automate the detection of explosive materials in both cabin and hold baggage. We have arrived. What they call Explosive Detection Systems (EDS) are in use in most of the world’s largest airports’ hold baggage screening operations, and airports in the U.S., Europe, and parts of Asia are actively deploying EDS for cabin baggage screening at the checkpoint. For the checkpoint, and from an airport resource perspective, EDS provides a tool to assist human operators in identifying suspicious items for secondary screening.
• Automated detection of prohibited items.  Machine learning and artificial intelligence (AI) have furthered efforts to automate the detection of non-explosive prohibited items (e.g. knives and guns) through shape recognition or from metadata.  Whereas several vendors have made noteworthy progress, actual deployments have been slow-going. Getting this done is the only way that operators will ever be able to achieve a fully automated screening process (automating explosives detection is only half of the puzzle).

You may be wondering why the automated detection of prohibited items has not kept pace with the automated detection of explosives?  We could write an entire article on this subject, but a few reasons include:

• Governments prioritize explosives (hint: they are more dangerous), and fund development accordingly.
• Governments have clearly defined explosive detection roadmaps (i.e. what to find, and in which order), and vendors know what to build.  No such roadmap exists for prohibited items (hence the point of this article).
• A higher-grade explosive detection capability will eventually open a new market (i.e. governments will mandate operators to buy it).  Governments are not so focused on saving airports money, so it is not as clear-cut whether they will mandate fully automated detection if it delivers the same security benefit and no more (even if it saves airports millions).
• Shape recognition tech has only come of age in the last 2-3 years.  If you ask a regulator today why there is not an automated detection roadmap for prohibited items, they will respond that the “tech hasn’t been there”. Chicken. Egg.
• People keep mixing up open architecture and automated detection. Open architecture is a hornet’s nest with all sorts of possible IP issues.  Automated detection of prohibited items is about vendors (or third parties) collecting loads of data and building solid algorithms (from shape recognition and/or metadata).

Back to the burning issue.

Operators: Governments know you are hurting, and they are willing to think outside of the box to ease the pain. This is your ticket to reduce x-ray operator review load by 25-50% in the next 12-24 months, with more honey to follow.

Vendors: Yes, I know you are killing yourself with testing and evaluation for explosives detection, and that you hardly have time to think about shape recognition (and you have no faith that regulators will build a testing and evaluation framework that works for it within the next 5 years!).  BUT! What if you had a clear detection list (that wasn’t the full prohibited items list)?  What if you knew that by meeting the requirement, an airport could use your tech to eliminate an x-ray operator with the same level of security, thereby creating a strong business case (i.e. they’d buy it tomorrow)?  New market.  New opportunity.  No more algo development to meet individual airports’ operational efficiency needs with little to no promise of a broader pot of gold.  The market will be clearly defined, and airports will love you.

Regulators: You have an opportunity to drive innovation to help the operators AND maintain and/or enhance security. It is time to define a clear automated detection roadmap and incentivize roll-out of the technology in the regulation. Doing this successfully will require collaboration between the U.S. Transportation Security Administration (TSA), the European Commission (EC), and relevant European Member States to:

• Measure and clearly define the efficacy of the human operator detection “system” in place at checkpoints currently to ensure that we’re defining a proper standard, moving away from the pointless “perfect algorithm” pursuit.
• Apply a risk-based approach to the threat list. We do not need to find everything on the list (baseball bats, really?) if we leave the operator on-screen for a certain percentage of the time. We use risk-based approaches in other areas; let’s do it here as well. Think of it in terms of pushing more and more into the auto-clear and auto-alarm buckets.  There will always be items in-between. Fear not.
• Harmonize the roadmap between the U.S. and Europe, write it down, give it to the vendors, and commit to a regulatory change that will open the market. Do it quickly. The industry is suffering.
• Design a third-party testing framework with harmonized government oversight practices between the U.S. and Europe and roll it out yesterday. The tech is ready, friends. Be a change agent.

In this instance, lack of a clear standard and a strong business case for the operators created through the regulation has delayed progress. Coronavirus or not, “allez regulators, let’s go!” And, vendors, do not disappoint please. The airports are counting on both of you. #ittakesavillage

By Anne Marie Pellerin, Founder & Managing Partner, LAM LHA