Accelerating Defence with GPGPUs: Unleashing the Power of AI and ML in Military Operations
From Big Data to Battlefield Dominance - How GPGPUs are Transforming Military Capabilities
Content overview:
- Evolution of big data
- CPU vs. GPU: What's the Difference?
- The Evolving Role of GPGPUs in Critical Military Intelligence
- GPGPUs in Unmanned Vehicles
- Advantages
- Ministry of Defence’s (MoD) Defence Artificial Intelligence Strategy
- A Real-world example of AI and ML during military operations
- Exercise Spring Storm
- Rugged Solutions for UGV, UAV and UUV Systems by Aitech
- Small Form Factor Supercomputer for Unmanned Aerial Vehicles (UAVs)
- Small Form Factor Supercomputer for Unmanned Ground Vehicles (UGVs)
- Small Form Factor Supercomputer for Unmanned Underwater Vehicles (UUVs)
Evolution of big data
Every day, we generate, create, capture, copy, and consume an immense amount of data worldwide, an amount that would only be unimaginable just a couple of decades ago. As the amount of data increases, so does the need for processing capabilities in embedded electronic systems.
This explosion of data generation in recent years has created a challenging landscape for traditional computing architectures, which rely heavily on central processing units (CPUs). These systems are finding it increasingly difficult to meet the increasing demands of modern artificial intelligence (AI) and machine learning (ML) applications.
While AI and machine learning tasks require processing vast datasets at high speeds, standard CPUs are proving to be a bottleneck. Designed for linear processing, they lack the necessary performance for these demanding workloads, creating a clear need for more sophisticated computing approaches.
CPU vs. GPU: What's the Difference?
When it comes to CPU’s handling AI and ML, the problem comes from a basic mismatch in how things are designed. Like Figure 1 shows, CPUs are great for running tasks one at a time and dealing with complex control functions. However, they fall short when it comes to the parallel processing needed for the matrix operations and vector calculations that are key to AI and machine learning (ML) algorithms.
General-purpose graphics processing units (GPGPUs) are designed with thousands of small, energy-efficient cores that work in parallel. These cores can run thousands of threads at the same time, which means GPGPUs can handle large datasets faster than regular central processing units (CPUs). This makes them an ideal solution for demanding tasks like training neural networks or processing data from sensors.
The Evolving Role of GPGPUs in Critical Military Intelligence
With GPU computing accelerating automation, especially for tasks requiring real-time decision-making, high-speed data processing, and improved safety.
The defence sector stands at the forefront of this, leveraging GPUs to drive modernisations on the frontline from MBTs (Main battlefield tanks) to vehicles that can carry out more autonomous tasks, such as unmanned aerial vehicles (UAVs), unmanned ground vehicles (UGVs), and unmanned surface vessels (USVs).
These systems operate in unique, unpredictable environments where split-second decisions are critical, requiring the parallel processing capabilities of GPUs to analyse vast sensor data streams, train complex neural networks, and execute AI-driven actions autonomously.
Whether conducting surveillance in hostile zones, detecting explosives, or coordinating swarm tactics, GPU-accelerated AI enables real-time threat classification, environmental interpretation, and adaptive responses. This reliance on GPU infrastructure aligns with the MoD’s AI strategy, which emphasises cutting-edge computational power to maintain tactical superiority. This is mirrored by the MoD's recent artificial intelligence strategy.
GPU Architecture: Powering Speed and Scalability
The unique design of GPGPUs offers several tactical advantages:
- Massively Parallel Processing: Multiple GPGPUs can be integrated into a single system, enabling scalable, high-efficiency computing for handling large datasets.
- Low Latency: Hardware optimisation minimises delays, ensuring real-time responses in dynamic environments.
- Bottleneck Reduction: Parallel architecture alleviates computational bottlenecks common in CPU-dependent systems, enhancing reliability in data-intensive missions.
Ministry of Defence’s (MoD) Defence Artificial Intelligence Strategy
The shift of embedded GPGPUs aligns with priorities outlined in the UK Ministry of Defence’s (MoD) Defence Artificial Intelligence Strategy, which emphasises the need for “fast, scalable, and secure compute power” to fully exploit AI across military operations (Section 2.2.2). The strategy highlights that seamless network infrastructures and state-of-the-art hardware are essential to enabling data to “flow seamlessly across sensors, effectors, and decision-makers,” thereby delivering strategic advantages in battlespace awareness and operational efficiency.
Specifically, the MoD’s vision for a “Digital Backbone”. A unified ecosystem integrating data, technology, and personnel relies on robust computing architectures like GPGPUs to meet real-time processing demands. For instance, GPGPU-powered AI models can process satellite imagery, radar signals, or unmanned vehicle feeds in milliseconds, enabling commanders to act on insights faster than opposing forces. Such applications exemplify how the military’s reliance on GPGPUs is not merely a technical upgrade but a strategic imperative to dominate in an era defined by data-centric warfare.
Examples of GPGPUs in Military Unmanned Vehicles
Unmanned Aerial Vehicles (UAVs) 'Drones'
Consider a UAV like the RAF’s MQ-9 Reaper (MQ-9A) a remotely piloted aircraft designed for long-range persistent intelligence, surveillance, target acquisition, and reconnaissance (ISTAR).
To achieve this, the aircraft is equipped with high-resolution cameras, infrared sensors, synthetic aperture radar (SAR), and LiDAR, generating gigabytes of sensor data every second. Traditional CPU-based systems would struggle to process this in real-time, delaying critical insights like identifying camouflaged targets or suspicious movements.
However, by leveraging General-Purpose Graphics Processing Units (GPGPUs), UAVs like the Reaper can rapidly analyse video feeds, radar returns, and infrared imagery using parallel processing. Computer vision algorithms accelerated by GPGPUs to enable real-time detection of hidden targets, anomalous activity, or changes in terrain.
This reduces reliance on raw data transmission and speeds up decision-making for operators.
Unmanned Ground Vehicles (UGVs)
Unmanned ground vehicles (UGVs) are currently deployed primarily for explosive detection, disposal, and reconnaissance in complex terrains or buildings. While most UGVs rely on teleoperation (remote control by human operators) some can execute basic autonomous tasks, such as point-to-point navigation, with minimal human intervention.
Like other autonomous vehicles, UGVs must overcome particular challenges within the battlefield such as navigating streets with rugged terrain, determining civilians from opponents, and operating in bandwidth-constrained or GPS-denied environments.
To handle these demands, today's UGVs leverage General-Purpose Graphics Processing Units (GPGPUs) to allow advanced autonomy.
GPGPUs process vast streams of data from cameras, LiDAR, and radar in real time, fusing these inputs to construct a dynamic 3D map of the environment. This parallel processing power supports computationally intensive tasks like simultaneous localisation and mapping (SLAM), object recognition, and adaptive path planning, allowing UGVs to autonomously navigate obstacles, avoid threats, and adjust to disruptions like smoke or signal loss.
Unmanned Surface Vehicles (USVs)
Maritime defence also presents an intelligent use case for GPGPUs. Unmanned Surface Vessels (USVs), used for patrolling coastal waters, leverage GPGPUs to integrate data from sonar, satellites, and thermal imaging. This enables them to detect submarines and identify illegal cargo shipments.
By processing AI models locally, USVs reduce their dependence on vulnerable communication links, such as soldiers on foot, providing uninterrupted mission operations even in challenging environments.
What are the advantages of using Artificial Intelligence (AI) and Machine Learning (ML) in military systems?
Although, this is just a glimpse of the significant role AI and ML play in modern military applications. Below are more examples of real-time processing demands, enabled by architectures like GPGPUs:
- Autonomous Navigation: AI-driven pathfinding and obstacle avoidance via LiDAR and camera data.
- Real-Time Intelligence: Rapid analysis of imagery and sensor data for signals intelligence (SIGINT) and intelligence, surveillance, and reconnaissance (ISR).
- Swarm Coordination: Efficient management of data for synchronised drone swarms and formation flights.
- GPS-Denied Environments: Simultaneous Localisation and Mapping (SLAM) algorithms enable navigation without satellite reliance.
- Hazard Detection: AI-powered recognition of improvised explosive devices (IEDs) and terrain obstacles.
- Autonomous Mobility: Real-time processing for route optimisation in complex environments.
- Mine Detection: High-speed processing of sonar and visual data for minehunting missions.
- Maritime Surveillance: Enhanced situational awareness through advanced sensor fusion.
A Real-world example of AI and ML during military operations
Exercise Spring Storm
In May 2021, the British Army conducted an operation which utilised AI for the first time. Soldiers from the 20th Armoured Infantry Brigade trialled a Machine Learning engine designed to process vast amounts of complex data, including environmental and terrain information. This engine demonstrated the potential to significantly reduce planning time compared to human teams while producing results of equal or higher quality.
The trial showcased the ability of AI to rapidly process large datasets, providing commanders with superior information during critical operations. By transferring the cognitive burden of data processing from humans to machines, the trial highlighted the potential for AI to revolutionise decision-making.
Developed in partnership with industry, the Machine Learning engine employed automation, smart analytics, and supervised learning algorithms. This AI capability can operate independently. By saving significant time and effort, it provides soldiers with immediate planning support.
Rugged Solutions for UAV and UUV Systems by Aitech
Pioneers in the advanced development of GPGPU-based AI applications, cybersecurity, and reliable digital connectivity across all domains of defence, Aitech provides state-of-the-art AI GPGPU-based boards and small form factor (SFF) AI systems specifically designed to excel in airborne, ground, and maritime applications, meeting stringent survivability requirements.
Adressing the demand for AI for various functions, Aitech's rugged GPGPU product line offers leading-edge solutions for video and signal processing, as well as accelerated deep-learning for the next generation of autonomous vehicles, surveillance and targeting systems and electronic warfare (EW) systems.
Small Form Factor Supercomputers for:
- Unmanned Aerial Vehicles (UAVs)
- Unmanned Ground Vehicles (UGVs)
- Unmanned Underwater Vehicle (UUVs)
- Unmanned Surface Vehicles (USVs)
UGV, UUV, USV and UAV
A230 Vortex, GPGPU Fanless Small Form Factor (SFF) Supercomputer
NVIDIA® Jetson™ AGX Orin System-On-Module
The A230 Vortex stands out as the most powerful Rugged GPGPU AI supercomputer, ideally suited for distributed systems, available with the powerful NVIDIA Jetson AGX Orin Industrial System-on-Module. Its Ampere GPU features up to 2048 CUDA cores and 64 Tensor cores, delivering up to 248 TOPS and ensuring remarkable energy efficiency for AI-based local processing right alongside your sensors.
In addition, the system includes two dedicated NVIDIA Deep-Learning Accelerator (NVDLA) engines, tailored for deep learning applications.
UAV and UGV
A178 Thunder, GPGPU Fanless Small Form Factor (SFF) Supercomputer
NVIDIA® Jetson™ AGX Xavier System-On-Module
Aitech's A178 Thunder supercomputer provides generous AI for military drones, as the systems are ruggedised to military specification and run on parallel NVIDIA general-purpose computing on graphics processing units (GPGPU).
Combining rugged durability with NVIDIA® Jetson AGX Xavier’s advanced computer architecture, this ultra-compact system delivers 32 TOPS (INT8) and 11 TFLOPS (FP16) performance through its 512 CUDA cores and 64 Tensor cores—enabling real-time processing of massive SIGINT datasets and AI-driven decision-making at the edge.
UAV and UGV
A179 Lightning, Rugged Fan-less Xavier™ NX AI Supercomputer
Intel ® Core® i7 Processor
The A179 Lightning is the smallest and most powerful Rugged fan-less AI supercomputer based on NVIDIA Xavier™ NX, brings AI performance to the edge, available with the powerful NVIDIA Jetson Xavier™ NX System-on-Module. Its Volta GPU with 384 CUDA cores and 48 Tensor cores reaches 21 TOPS INT8 at a remarkable level of energy efficiency, providing all the power needed for AI-based local processing right where you need it, next to your sensors.
Two dedicated NVDLA (NVIDIA Deep-Learning Accelerator) engines provide an interface for deep learning applications. With its compact SFF size, the A179 Lightning is the most advanced solution for AI, deep learning, and video and signal processing for the next generation of autonomous vehicles, surveillance and targeting systems, EW systems, drones, wearable and many other applications.
UGV and UUV
A172 | Small Form Factor High Performance Embedded Computer
Intel ® Core® i7 Processor
The A172 is a low profile, rugged high performance embedded computer (HPEC) that provides exceptional design flexibility in a low power, high performance computing system. Developed around a standard Type 6 Com Express module, the A172 features an industry standard pinout and the ability to support multiple processor options for easy system integration as well as cost-effective technology insertion upgrades.
Weighing under 5 lbs (2.25 kg) with dimensions of only 10.24 x 7.09 x 1.8” (260 x 180 x 46 mm), the A172 maximizes SWaP-C for applications with limited space requirements.
It also features a flexible architecture for upgrading the CPU to the next generation, as well as expansion options such as mini-PCIe cards and up to 1 TB of removable storage. The A172 includes multiple I/O interfaces that make the rugged, Intel Core I7-based unit ideal for high impact, high performance military applications. These include manned and unmanned fixed- and rotary-wing airborne (UAV) platforms.
About the Author
Matteo Magnifico
Head of Sales Europe | Aerospace, Space and Defence at APC Technology Group
As a member of our Aerospace and Defence division at APC Technology Group, Matteo Magnifico is an expert in European AeroSpace, Defence and Industrial Markets.
Contact us
Over 40 years of expertise supporting defence manufacturers and military primes with highreliability and MIL-STD technology solutions for naval, land and avionic applications.
As a leading distributor to over 100 specialist technology manufacturers, APC’s team of experts advise and supports the design-in and supply of a comprehensive range of electronic solutions. From GPGPU components to rugged computing systems – our teams help to drive forward the defence technology of tomorrow