How artificial intelligence is improving automotive manufacturing standards

Posted on 27 Nov 2019 by The Manufacturer

Amir Hever, CEO of UVeye, recalls the company’s beginnings and explains how its range of AI-based products is setting the future standard for automotive manufacturing.

UVeye is a provider of solutions for the automatic external inspection of vehicles. With roots in the security and defence sector, the company’s threat-detection products have already been deployed in numerous high-security facilities around the world.

More recently, UVeye has extended and adapted its product suite for the wider automotive market. Not only is the company’s technology targeted at improving quality standards in garages and dealerships, but also in vehicle manufacturing plants, particularly in end-of-line inspection.

Automotive Conveyer in factory Production Line - image courtesy of Depositphotos.

Image courtesy of Depositphotos.

It was after a visit to the premises of a government agency in Tel Aviv that my brother, Ohad, and I had the idea for UVeye – short for Under Vehicle eye.

When we arrived, a security guard stopped us and laid down on the ground next to our car. I asked him what he was doing, and he told me that he was looking to see if I had anything hidden under the vehicle.

“Do you see anything?” I asked. “Nope, I can never see anything,’’ was his swift reply.

From the resulting conversation, I discovered that, even in a country like Israel, where the highest safety standards are expected, and even in a world where digital technology has made leaps and bounds in recent years, many vehicle inspections still consist of rudimentary visual checks.

In the hope of offering an alternative method, in June 2016 we decided to set up UVeye with the aim of utilising high-resolution cameras and AI-based software solutions to take external vehicle inspections into the digital era.

This article first appeared in the November issue of The Manufacturer magazine. Click here to subscribe

The power of three

Naturally, developing technology that could examine the undercarriage of a vehicle was a good place to start, given our experience at the government agency.

However, after introducing the system as a tool in the security sector, we also saw the potential to take this technology further and bring it to the automotive industry.

Just as the technology could be used to detect threats on a vehicle, so too could it be used to detect faults, maintenance issues, and cosmetic damage.

A key area for its application is during end-of-line inspection in an automotive manufacturing plant, where the vehicle is visually inspected for imperfections.

CROP - Businessman on blurred background using digital artificial intelligence icon hologram 3D rendering - image courtesy of Depositphotos.

Image courtesy of Depositphotos.

With AI-based technology, it’s possible to increase the efficiency, objectivity and accuracy of work on vehicle production lines, while enhancing safety and enabling a higher volume of work with the same amount of resources.

By detecting faults at an early stage, we can prevent a potential breakdown and reduce maintenance costs over the lifetime of the vehicle. These faults might include loose bolts, incorrectly routed cables, damage to paintwork or underinflated tyres, to name a few examples.

What’s more, with manual checks, manufacturers not only risk overlooking faults on their vehicles, but also waste time that could be more productively allocated elsewhere in the factory.

An intelligent AI-based system greatly enhances speed and efficiency, improving the flow of vehicles through and out of the plant.

With all of this in mind, we expanded the breadth and capabilities of UVeye’s technology to other areas of a vehicle’s exterior, such as the tyres and bodywork.

Today, we have three product offerings: Helios, Artemis, and Atlas.


Helios is the undercarriage inspection system, which comes in stationary or portable iterations.

UVeye’s Helios undercarriage inspection system - image courtesy of UVeye.
UVeye’s Helios undercarriage inspection system – image courtesy of UVeye.

The hardware comprises five high-resolution cameras in its stationary version and three in the portable version, which capture images from multiple angles to ensure that no spot on the vehicle goes uninspected.

Helios produces a high-resolution image within three seconds of a vehicle driving over the hardware – at speeds of up to 30km/h – and a full analysis within 10 seconds. The system provides the user with a detailed view of the complex componentry on a vehicle’s undercarriage.

UVeye’s deep learning algorithms then process the image to detect anomalies that would otherwise go unnoticed by the human eye.

The technology is able to accurately identify what the individual parts of the undercarriage look like under a wide range of different conditions, such as lighting, stages of wear and tear, and moisture.

If an anomaly is detected, the Helios system automatically alerts the employee at the manufacturing plant and directs them to the exact spot where the anomaly was found.

This makes it easy for them to take action, while reducing the learning curve and other factors like fatigue or stress.


UVeye’s Artemis tyre inspection product - image courtesy of UVeye.
UVeye’s Artemis tyre inspection product – image courtesy of UVeye.

Artemis is UVeye’s tyre inspection product. The system comprises two tyre scanners that stand at the side of the vehicle while it drives past at speeds up to 20 km/h.

In a matter of seconds, Artemis reads and recognises the tyre brand, markings, and technical specifications, as well as crucial safety-related data such as tyre condition, pressure, abrasions and to cross-reference the pressure of each tyre to the manufacturer’s standards and measurements, and report any incorrect pressure levels.

The system can even provide a comparison across all of the vehicle’s tyres to determine any irregularities. Presented with this wealth of information and a high-resolution image that highlights any faults or anomalies, technicians at the manufacturing plant can then take the necessary steps to repair or replace the vehicle’s tyres.

It can also bring to their attention any repetitive issues that could be indicative of a larger problem in the production process that must be corrected.


The 360° full-body scanner, Atlas, completes UVeye’s portfolio and is used to detect dents, scratches or other cosmetic issues on the vehicle’s upper bodywork.

Atlas – UVeye’s 360° full-body scanner - image courtesy of UVeye.Atlas – UVeye’s 360° full-body scanner - image courtesy of UVeye.
Atlas – UVeye’s 360° full-body scanner – image courtesy of UVeye.

Atlas looks similar to the large light tunnels found in vehicle factories, taking the form of a bright arc that encircles the vehicle. However, it is far more compact and incorporates the same deep-learning algorithms as Helios and Artemis.

With these, the system can automatically detect scratches as short as a few millimetres, in addition to any broken parts.

Unlike conventional methods, vehicles can be driven or moved on a conveyor through Atlas at up to 20km/h, allowing an uninterrupted flow across the production line. Only if a fault is found does the line need to be stopped.

Working together with Helios and Artemis, the three systems are capable of producing a full 360° scan of the vehicle, so that no fault goes unnoticed.

How does the technology work?

UVeye’s sophisticated piece of technology solution is based on deep learning and image processing. It uses advanced computer-vision algorithms to provide a detailed visual analysis of a vehicle’s exterior and highlights any areas of concern.

Our advanced algorithm meets the challenge of automating anomaly detection for new and unfamiliar ‘first pass’ vehicles. The technology analyses each vehicle part separately, detecting anomalies within seconds – but crucially without the need to reference a previous scan or undercarriage image provided by a vehicle manufacturer.

This allows us to catch issues not simply because it failed to match an ideal image of what the manufacturer thinks their vehicle is supposed to look like, but because we have trained our algorithm to truly understand what the parts of the vehicle look like under a range of conditions.

Everything from the lighting, moisture levels and other natural conditions can change how an image may appear, thus requiring us to train our deep-learning systems to detect and identify anomalies regardless of the situation in the field.

A key aspect of our technology is our ability to compile the images using an area patching methodology that ensures that no part of the vehicle is missed as it moves speedily through the scanners.

Investment for the future

UVeye co-founders Ohad (left) and Amir Hever
UVeye co-founders Ohad (left) and Amir Hever

We recently closed our Series B of funding, raising $31m, led by Toyota Tsusho, Volvo Cars, and WR Berkley.

The investment we’ve received from leading automotive strategic partners is an important signal that we believe paves the way for UVeye to become the standard of automotive inspection and safety.

The latest round of funding enables us to continue enhancing our product suite and further expand our international footprint as the emerging global standard for automatic vehicle inspection.