From man to machine: Thales DigitalCrew and the future of vehicle optics

Modern armoured fighting vehicles carry a significant suite of optics. From commander’s sights to situational awareness systems, there is a lot for the vehicle crew to process. So, Calibre Defence spoke with Stewart Macpherson, Head of Digital Strategy at Thales UK. The aim was to learn more about how the company is addressing the need to process this imagery, and its DigitalCrew.

By Sam Cranny-Evans, editor of Calibre Defence, published on March 19, 2026. 

For most of the history of the modern tank, the crew’s vision has been quite limited. If a commander wanted to know more about the world around him, the most effective option was often to take a look. Exposing his head and chest, he would open a hatch to look and listen, trying to understand the evolving battle. The periscopes arranged around that hatch would not give the same view or understanding — they might be murky, and require the commander to move his head or body to switch between them. A commander or gunner’s sight provided magnification, and over time evolved to bring thermal imaging to the frontline. Both tools increased the tank’s lethality and survivability considerably. And yet, when all else fails, the commander is still inclined to open a hatch and observe with the ‘Mark 1 eyeball.’

Tank commanders have always suffered heavy casualties because of this. Most notably, Israeli officers — who tend to lead from the front — are regularly targeted as they appear from the turrets of their Merkavas. But whether it was periscopes, a commander’s sight, or the Mk 1 eyeball, it has always been the human brain that makes sense of the battlefield. And this is where the world of armoured fighting vehicles is set to change.

DigitalCrew: the tech stack

When Calibre Defence last spoke with Stewart Macpherson, the Head of Digital Strategy at Thales UK, it was about the TrueHunter Commander Sight. This is to be deployed on the Challenger 3 tank being built for the British Army. At that time, Stewart made a few references to something called DigitalCrew, so this interview went a little further. But let’s start with an explainer, what is DigitalCrew?

In a nutshell, DigitalCrew is a collection of Generic Algorithm Building Blocks designed to support vehicle crews. It does this by helping them to find and identify targets through their sights. Stewart explained those algorithms as follows: 

• Object Detection: A mathematical algorithm — not AI or machine learning — that detects moving or hot objects without requiring training.
• Object Tracking: Locks an aiming mark onto a selected object, with the weapon system automatically following it. Particularly useful for remote weapon stations tracking fast-moving targets like drones, which are difficult for operators to follow manually.
• Multi-Object Tracking: Where multiple objects appear in the field of view, the algorithm automatically assigns, tracks, and prioritises each one simultaneously.
• Object Classification: Automatic recognition of objects of interest within the field of view dependent on the training data provided to the ingested by the model; people, vehicles, drones, sub-sets of the classes.
• Video Combination: Stitches multiple sensor feeds together — for example, rotating a sight 360 degrees to produce a panoramic image. Object classification algorithms can then be overlaid, using a neural network trained to identify people and vehicles and assign a confidence level.
• Mission Support Tools: Currently at a low TRL, these tools use the outputs of the above algorithms to provide higher-order support to the operator. Potential capabilities include passive ranging, threat assessment, and pattern of life analysis.

But the project started six years ago, and the reasons for that are worth exploring. 

DigitalCrew: the back story

Thales is a big company with a lot of specialisms. But the UK entity can proudly trace its history back to a company called Barr and Stroud. Based out of Glasgow, Barr and Stroud provided optical range finders for the British Army in World War 1, as well as the country’s first submarine periscopes. It has stayed involved in the optics world for defence ever since. The commander’s sights for the Ajax reconnaissance vehicles, and the TrueHunter sight on Challenger 3 are both made by Thales UK. 

Barr and Stroud provided Thermal Observation and Gunnery Systems for Challenger 1, making a definitive contribution to the Gulf War. Thales produced the Battle Group Thermal Imaging sights for the UK’s Warrior IFVs and Scimitar reconnaissance vehicles. And it went on to produce TOGS II for Challenger 2. So, it is safe to say that Thales UK knows what it is doing when it comes to vehicle optics. 

• Challenger 3: Inside the TrueHunter Commander Sight with Thales – Calibre Defence 
• Challenger 3: the heart of the Army’s warfighting capability – Calibre Defence 
• Challenger 3: First crewed live firing trials completed – Calibre Defence 

But in 2020, the company set out to augment the humans inside armoured fighting vehicles. Referring back to the sensor proliferation, Stewart said: “We have to develop a way for the machine, the algorithm to identify critical information and assist the human. For instance, can the machine track an object rather than a human, or keep an aiming mark on a drone? Or alert the crew to a suspicious object at range?” 

This is where DigitalCrew comes in; combining a deck of algorithms, it is hoped to alleviate the cognitive load of a combat crew. “We are sensor experts, and we are trying to take that knowledge to optimise the algorithms for those sensors,” Stewart explained. “Algorithms can work well in what we call ‘sunny day scenarios’ but we have experience of how these sensors perform in all environmental conditions. Day and night, at all ranges and when sensor noise may become a factor,” he added.

Ajax and its cameras

The project can trace its roots back to Ajax, which Stewart sees as a demonstration of sensor proliferation. “Ajax has 28 sensors on it, two for each sight and the local situational awareness cameras have 2-3 cameras per station. That’s a lot of data, and for a human operator it would be overwhelming,” he said. 

To help the Ajax commander and gunner, Thales developed an Automated Target Detector (ATD) and Automated Target Tracker (ATT).  ATD and ATT are logic based, this means they are mathematical algorithms rather than artificial intelligence (AI). 

The technology at the core of these algorithms is classical computer vision. In a nutshell, the algorithm looks for specific mathematical signatures, such as a high-temperature pixel cluster against a cold background. Following a strict “If X, then Y” logic, these algorithms can help a commander make sense of the imagery they are receiving. AI can be used, but it is impossible to explain why or how it arrives at the conclusions it does. 

“ATD works on object detection, it will find hot or moving objects. But ATT is more complex, it’s actually a cluster of algorithms. If you want to keep an aiming mark on something, you select it and the weapon system will follow it.” Stewart went on to explain that DigitalCrew groups these algorithms together, adding machine learning for object classification. 

SWaP-C and the computing problem

Anything built for military uses will likely have to address an element of SWaP-C, which stands for Space, Weight and Power-Cost. A satellite represents one of the more acute SWaP-C challenges for a design team. Every hundred grams of added weight to a satellite will increase the cost of launch exponentially, for example. SWaP-C also applies to the inside of a tank or armoured vehicle, most of which are very small inside. 

However, Stewart offered an additional vector to the SWaP-C paradigm, computing. “For DigitalCrew, the idea was to make these basic building blocks that can run on the kit that the army already has. So, we have optimised some of them to run on x86-based CPUs.” An x86 CPU is basically any computer that is not a Mac. It uses the system architecture developed by Intel in the 1970s and is backwards compatible. In theory – code written in 1978 would still run on a modern laptop. And while an x86-based CPU is not necessarily limited in power, it lacks the parallel processing and complex mathematical functions needed to make large AI models work. 

“It’s a very big technical challenge to make algorithms lightweight, and there are points at which you cannot do that. Machine learning will struggle to get operationally relevant framerates on a CPU,” Stewart added. Emerging techniques like quantizing and minifying (yes, it’s a real word) are emerging to squash AI models, allowing them to run on smaller chips. But this does not change the fact that a computer in an armoured vehicle might be a decade old. So, algorithms need to be able to run on those systems, or a designer needs to find the SWaP-C to make a new computer it in the vehicle. But this too comes with its challenges. 

The future of edge computing in defence

“This problem of CPU/GPU capability will only get worse. Nvidia has said that they will only support GPUs for five years, and then you need to buy a new one. We need to think about that going forward, and consider who is responsible for upgrading that GPU,” Stewart said. This will create friction within defence procurement. Changes to computing can require a system to be requalified and tested to ensure it meets design standards. It’s a slow and expensive process, but Stewart offered a potential solution. 

• Second Front and Valarian partner for secure software deployment 

“All of our sighting systems we supply for the British Army (including Boxer) come with a sight processing system. It’s essentially a ruggedised PC that we should see as edge computers, rather than a box attached to the sight.” In short, there is unused computing power attached to a vehicle’s sight that can be put to use. DigitalCrew is of course an option, and Thales is exploring ways to make the system most valuable for its users. But Stewart was also keen to emphasise that the system is designed to be vendor agnostic. Other companies can develop models and algorithms to run as part of DigitalCrew, providing users with options for future upgrades. 

It certainly has its appeal as new capabilities can be rapidly added to a vehicle. Helsing and General Dynamics cooperated in 2024 to send object classifications as metadata over the Bowman network, for example. That upgrade, which was applied to Ajax, was primarily achieved through software changes. Stewart added that interoperability is one of the founding principles of DigitalCrew. “Ajax should be able to talk to a Leclerc or a German UAV. Because that is what could happen…Digital Crew is just a framework for how to specify and integrate algorithms into the video chain of a sensor on the edge. This provides the lowest latency path to the existing sensor.” 

Calibre comment: DigitalCrew and 20-40-40

The British Army is looking to develop a hybrid force with 80% of its combat power coming from uncrewed platforms. This includes munitions, drones, and UGVs, all of which will be launched and controlled from crewed platforms. This represents an integration challenge – to say the least. Some systems like Cobalt from Arondite, which was recently contracted by the MoD, could offer a solution. But it is likely that vehicle crews will need to command uncrewed platforms from their turrets. In those scenarios, the crew would likely be faced with a greater volume of data. An additional metadata feed from UGVs pushed ahead of a force, for example. Here it seems that systems like DigitalCrew could help support those types of concepts. However, there is a long way to go before that kind of integration can be achieved. From procuring the uncrewed systems, through to making sure they can talk to other platforms, let alone be controlled by existing crews without impacting their performance. There are many different challenges that will need to be addressed before hybrid fleets can be fully realised. 

If you would like to read more on the UK’s efforts to develop a hybrid force, have a look at the articles linked below: 

• Project ASGARD; the British Army’s path to doubling lethality – Calibre Defence 
• Haraka Storm: ARX and Helsing close a robotic recce-strike loop – Calibre Defence
• Arondite to give 16 Air Assault more punch with Cobalt – Calibre Defence  

The lead image shows a Boxer armoured vehicle fitted with the Thales TrueHunter gimbal sight. Credit: Thales UK.