Harnessing The Power Of Automation For Vehicle Inspections by Amir Hever

Every aspect of physical security is being improved by the intelligent use of technology. In the current climate of constant threat from terrorism even the checking of vehicles entering high security sites is changing. Less reliance on a mirror on a stick and more of this sort of tech? What am I talking about? TPSO asked CEO of Uveye, Amir Hever, to explain…

In security, everyone wants their solutions to be good, fast, and cheap. Attaining all three, however, is far more difficult to achieve. This is doubly true when it comes to approaching the challenge of vehicle inspections.

Legacy practices for inspecting vehicles, such as the use of mirrors, provide neither the accuracy or speed that are necessary to contend with the amount of vehicles that need to be inspected. In response to this need, the security industry has witnessed the rise of automated solutions that can quickly and accurately scan vehicles for threats.

This article will break down the challenges posed by vehicles in the hands of attackers, assess the issues facing security professionals in performing their vehicle inspections, and lay out some of the most promising technological solutions available for making these inspections more comprehensive and efficient.

The Threats Posed by Vehicles — Understanding the Risks

Vehicles have long posed an added risk so long as there has been a way to hide illicit items on, in, or under them. Vehicle-borne improvised explosive devices (VBIEDs aka car bombs) have a long history of use by various non-state actors like the IRA and Red Army factions in the 1970s and 80’s.

In recent years, we have seen the vehicle used more by asymmetric groups and lone wolf-type actors to carry out attacks, often with significant effect. Attackers belonging to the Islamic State group, or inspired by them, have used vehicles in attacks in different forms throughout the globe.

Vehicle threats can be broken down into three basic categories.

Car Rammings

First is their use as a ramming device, driving through crowds to cause casualties and instill panic. These types of attacks are exceedingly difficult to prevent as they require a determined individual who takes the decision to carry out such an attack and a vehicle.

No additional illicit materials, training, or networks are needed for an attacker to be successful, just the motive and opportunity. They can cause considerable casualties if they strike at a public space like Christmas festivities and the like.

The best protection against these types of attacks is generally strategically placed bollards or other barrier types where possible. Unfortunately the options for detecting and preventing these ramming attacks are limited.

Weapons Smuggling

Vehicles provide an ideal mode of smuggling weapons. Attackers can move larger quantities of weapons than they could on their persons, avoiding detection through subterfuge.

Weapons can be hidden inside of doors and chairs, or even stuck under the vehicle itself, which makes it more challenging for security professionals to uncover as they often lack the necessary visibility to detect them quickly and accurately. Failure to detect weapons hidden in vehicles at a checkpoint may allow adversaries to bring weapons inside the perimeter and carry out their attack.

Vehicle-Borne Improvised Explosive Devices (VBIEDs)

The deadliest of the three categories, vehicles can be used by adversaries to carry out mass casualty attacks.

The U.S. Department of Homeland Security estimates that a compact car can carry up to 500 lbs of explosive materials, while a delivery truck can reach upwards of 10,000 lbs, impacting an area of over 5,000 ft. Along with any additional items like bolts and nails, the body of the vehicle itself can serve as shrapnel in the attack to increase its deadly effectiveness. By contrast an explosive-laden vest is likely to contain somewhere in the 20 lbs range that while deadly, poses a significantly lower consequence on our threat model.

VBIEDs are additionally challenging for security teams as they are difficult to identify from a distance. With the exception of the “Mad Max-esq” armoured constructions made by ISIS in Syria and Iraq for use on the battlefield, VBIEDs are meant to blend in with the public. This means that they are able to approach their target undetected, often until it is too late.

Mitigating the risks of VBIEDs requires a set of planning and operational practices:

  • Creating distance between the perimeter and likely targets like embassies, government offices, and other sites where there is a higher probability of VBIED use.
  • The use of physical barriers like bollards, wedges, and K12-tested fences that can prevent a vehicle from crashing through the perimeter.
  • Inspect vehicles as they enter the perimeter, checking them for explosives

Security teams can perform visual inspections of a vehicle’s storage areas — boot, cabin, or the back of a truck —  to identify gas canisters or other explosives, but this requires them to approach the vehicle, thus putting them at risk.

Challenges to Vehicle Inspections

Implementing a policy and infrastructure for proper vehicle inspections is not without its challenges.

The current geopolitical situation has raised the profile of security, leaving little room for failure. At the same time, security teams must contend with larger scales of vehicles at sites like border crossings, energy facilities, military bases, seaports, and more.

At the same time as security is put at a top priority, the increase in checks results in additional friction that can slow movement through checkpoints to a crawl.

In the case of commercial sites like seaports and border crossings, slow inspections have a direct impact on the number of vehicles that can pass through every day. This in turn correlates to productivity and profitability.

For more sensitive security sites like an embassy or hotel in a higher risk area, bottlenecks at the entrance to the perimeter can create a target for attackers, leaving those waiting for inspection vulnerable outside the protection of the perimeter.

A security team’s capabilities are limited in what they are able to accomplish using manual methods. Inspections that look under the vehicle are often less than effective with operators either using mirrors that only show a limited view of the undercarriage. In some cases a person can go under a vehicle for a closer look but this too takes time. Results from both methods are difficult to track well and are vulnerable to human errors.

How Automation Empowers Security Teams to do More

We know that the vast majority of traffic that passes through our checkpoints is likely to be benign. Our concern is that a number of wolves will try to pass themselves off in sheep’s clothing, counting on the pressure being placed on a security team to keep the flow of traffic moving through to limit the level of inspections.

Quite often this translates into security teams only performing random checks of selected vehicles. This may mean that one out of every few cars are inspected while the others are let through without any real accounting. The vehicles that are selected undergo either an intensive, time consuming check without any real indication that there is a threat or an inspection that is more cursory and unlikely to detect objects that are well hidden.

All of these scenarios as laid out are less than ideal to say the least. What is required is for security professionals to have increased visibility and the ability to quickly understand if a vehicle requires a more intensive inspection.

This is where automation and inspection technologies are able to step in and provide solutions, tackling the challenge from a variety of angles to produce a more comprehensive picture.

Top 3 Vehicle Inspection Technologies to Follow

Of the many advances in inspection technologies in recent years, these are three of the most relevant that should interest security professionals in terms of vehicle inspections.

Thinking about vehicle inspections, we can break down the process into three stages: identify, inspect, and alert.

Identify — LPR and Facial Recognition

When a vehicle pulls up to our checkpoint, the first action necessary is to identify who it is. License plate readers and facial recognition solutions play an important role here in telling us if the vehicle and passengers are who they say they are.

When connected with a database of vehicles and personnel, these systems can identify and compare the captured images to authenticate. Sites such as secure facilities that have authorized persons coming and going on a regular basis are an ideal use case for these types of solutions.

Inspect — X-Ray and Under Vehicle Scanning Systems

Especially in use cases like border crossings and ports that see lots of vehicles passing through every day, scanning solutions must be scalable.

When it comes to cargo scanning, x-ray technologies allow authorities to quickly capture images of loaded vehicles, breaking down the different items by elements for detection of illicit materials. With scans taking just a matter of minutes, security teams can process a significant number of trucks with a high level of confidence.

The undercarriage — as previously noted — is notoriously difficult to inspect, leaving plenty of room to hide items down below. Under vehicle scanners with high-resolution imaging are able to capture pictures of moving vehicles as they pass at the checkpoint, reducing friction without lowering standards of inspection.

Alerting — Using Artificial Intelligence to Alert on Anomalies

While high quality images can be helpful for security teams in identifying potential threats, areas like the undercarriage can still be difficult to deal with. Every car looks a little bit different, conditions may vary by the day, and not every threat looks like a cartoon bomb with a ticking clock.

This is where AI like deep learning and machine learning are playing an important role in helping security teams to identify threats. As algorithms are used in constantly scanning vehicles, they are learning what a vehicle is supposed to look like. This learning is not limited to specific vehicles by make or model, but across a wide range that allows the system to understand what is, and more importantly, is not supposed to be there.

Deep learning allows the system to detect anomalies and alert security teams to their exact location, leading to faster, more accurate inspections of relevant vehicles.

Recognizing the Limitations of Automation

Implementing automated solutions into vehicle inspections can do a lot to improve a security team’s accuracy and speed. At the same time, it is important to recognize its limitations.

Humans and canines are still critical parts of the inspection process at this point for seeking out illicit items in a suspicious vehicle, looking and sniffing around to uncover threats. They are the final stages of a process wherein a vehicle has been deemed in need of additional attention. Machines are getting smarter by the day but human security professionals will still be able to put into context situations and events in ways that digital methods are yet to catch up to at this point.

The real value of automated systems is to reduce the need for humans to inspect each vehicle by hand and eye. Instead it allows security professionals to direct their efforts at vehicles which have been flagged, allowing the rest to flow through while still having undergone inspection. 

In the three technologies included in this article, all are using automation to perform scans and, where possible, perform the work of detecting and analyzing the images for relevant, actionable data. This saves security teams the time and provides them with a valuable resource that they can utilize.

For security professionals looking to incorporate automated technologies into their existing workflow, it is important to think about these solutions not as replacements for their human teams but as an augmentation that allows their teams to do more with what they have. Security is done best when it works in layers, with managers using all the tools available to achieve their mission goals.

Amir Hever – CEO & Co-Founder Uveye

Amir Hever is the CEO and co-founder of UVeye, the deep learning computer vision start up that is setting the global standard for vehicle inspection with fast and accurate anomaly detection to identify issues or threats facing the automotive and security industries.

Hever co-founded UVeye in June 2016 along with his brother Ohad, who serves as the COO.

With an extensive background of over 10 years working in the field of computer vision, UVeye is Hever’s third venture. He previously held the position of VP R&D at Visualead which was acquired by Alibaba. There he created technology to improve interactions between people and brands through mobile innovation. Prior to Visualead, he worked with Correlsense, an enterprise-application performance-management (APM) company.