Case studies: artificial intelligence in the defence industry

The French defense and technology company Thales launched an AIbased tactical training and shooting analysis system at the Eurosatory 2022 defence exhibition. The offering includes training solutions for instructors and trainees and aims to ensure the safety of the training sessions.

Thales’ solution is a portable analysis and debriefing tool for the defence and security sectors that documents training session events and parameters such as shots, weapon data, videos, and positions. Low latency algorithms are made possible by AI-based tools, and the analysis tool instantly produces qualitative reports, scorecards, and indicators. It enables the instructor to receive objective data on each trainee and construct precise behaviour and group interaction analysis.

Thales asserts that the training solution meets the procedural and tactical needs of law enforcement agencies (LEAs) and special operations forces (SOFs).

LEAs and SOFs need to accelerate training for their teams on weapon handling, teamwork, and tactical operations and how to sustain abilities over time within the context of a dynamic international environment.

Existing training methods require conducting live sessions with real weapons in shooting ranges which are done a limited number of times a year. Thales’ solution records the training sessions and helps the trainer analyse the trainee’s behaviour and performance. 

The National Aeronautics and Space Administration (NASA) has unveiled spacecraft and mission hardware developed employing generative AI. These specialised components, known as evolved structures, are used in equipment such as astrophysics balloon observatories, Earth-atmosphere scanners, planetary instruments, and space telescopes.

The emergence of generative AI design could advance NASA's approach to conceptualising and testing components for upcoming robotic and human space missions.

During the design phase, an engineer instructs the system on what the part must do and how it will connect to other parts and identifies dead zones where material cannot be inserted. The AI system then uses computer-aided design (CAD) software to build the part per the pre-established specifications.

NASA states this process can construct a prototype item within a week upon completing the design, analysis, and production stages. The entire process, from concept to specification to finished product, is dramatically accelerated through the integration of generative design.

After receiving design feedback, an AI generates 30 to 40 iterations in an hour, simplifying the work of NASA engineers. NASA also uses generative design AI to automatically design wires and circuits for 3D printing.

The aerospace sector is among the most highly regulated industries, with components having significantly smaller tolerances for error due to the extreme applications in which they are typically employed. With few modifications, commercial-grade AI tools might be capable of creating components for crucial space missions.

NASA's design process begins with a prompt, employing geometric data and physical parameters as its inputs. The generative AI tool compresses and processes everything internally and independently, creating the design, analysing it, determining the manufactured product's viability, and performing corrective iterations quickly.

Evolved structure support struts designed for NASA's balloon-borne exoplanet-observing telescope EXoplanet Climate Infrared TElescope (EXCITE) mission are recent examples of aerospace components produced using generative AI-assisted design techniques. 

Israeli defence technology provider Elbit Systems has unveiled a novel combat drone called Lanius which can recognise and detonate explosives at specific targets using AI-assisted algorithms. The new drones are commonly referred to as "suicide" or "kamikaze" drones. These tiny AI-enhanced drones independently scan and identify buildings in metropolitan areas for potential threats.

Lanius is an adaptable and versatile drone-based loitering munition system intended for close-quarters combat in densely populated urban environments. The system can identify, categorise, and synchronise data transmissions with Elbit’s Legion-X autonomous combat networking platform while scanning and mapping buildings or locations of interest for potential threats.

Lanius drones also feature a modular payload, allowing them to be fitted with lethal and non-lethal effectors. The maximum take-off weight of a single Lanius is around 1kg, with a payload capacity of 5.3oz. It offers a seven-minute run time with energy sourced from a lithium-ion battery. Smaller drones can identify and categorise threats by scanning the battlefield and mapping its environments using AI-powered object detection. Lanius drones can also conduct improvised ‘ambushes’, by landing and positioning themselves to target key manoeuvre routes or points of entry into a room or building before detonating their payload once a viable target is detected.  

A loitering munition is a type of UAV containing an explosive warhead, designed to ‘loiter’ to identify and engage targets of opportunity beyond line-of-sight. Loitering munitions are frequently transportable and intended to give ground forces like infantry access to precision-guided weapons systems. Elbit’s Lanius platform employs the motor and airframe from racing quadcopter drones to optimise the platform for speed and agility.

The Lanius maintain the ‘human-in-the-loop’ safeguard, thus preventing the drone from engaging even armed targets unless the operator provides prior approval, underlining the ever-present concern regarding lethal autonomous weapons.

Users can operate Lanius independently or as part of a networked swarm, thus enabling armed forces to map the battlefield simultaneously using several tiny drones. Lanius is well suited to conduct various mission profiles for special forces, the military, law enforcement, and homeland security (HLS). 

The US Air Force Research Laboratory (AFRL) Munitions Directorate demonstrated the Team Eglin Weapons Digital Enterprise WeaponONE (W1) program at the Virtual Warfare Munitions Simulator in 2021. This was based on a model of the 24- hour air tasking order (ATO) cycle of the Gray Wolf collaborative swarming weapons system prototype.

To significantly increase Gray Wolf's capabilities, the virtual showcase combined many aspects of the W1 portfolio, including the Digital Twin Lab.

The digital twins run on high performance computer systems supported by AI algorithms to assess potential software improvements. After evaluating the best course of action, the intelligence is quickly sent to the actual weapons in-theatre, improving their effectiveness in near real-time or as early as the following 24-hour ATO cycle.

The presentation demonstrated how data from weapons in flight is gathered, integrated with data from the combat environment, and relayed to digital twins via the Advanced Battle Management System (ABMS). According to the AFRL Munitions Directorate, the Digital Twin Lab operates as a force multiplier, providing unprecedented flexibility and agility to modern weapon systems.

Digital twins are digital representations of physical assets, systems, or processes. The W1 program plans to progress its prototypes to "real" digital twins that can share data with their physical counterparts in both directions.

Its ongoing efforts in researching and developing ABMS, connected hardware, and simulation analysis will help to develop new capabilities that use digital twins to enhance the maintenance and procurement of equipment.