Artificial Intelligence learns, along with human operators, how to take control of the fight armored methods in the modern battlefield

From a series of speeches held during the “Artificial Intelligence Exploitation”, recently started by the Artificial Intelligence Consortium of the Naval Postgraduate School in Monterrey, US, aiming at exploiting and clarifying the Artificial Intelligence definition, the benefits and risks associated to its use, as well as the significance of its utilization in the military field, I found out that, in fact, the AI field was founded back in 1956, during the study of calculating machines construction that could accomplish intelligence tasks.

Stimulated by enthusiasm, affected by too optimistic and exaggerated promises, this field has already gone through three development waves, as I was previously mentioning in an article (“Handcrafted Knowledge”, “Statistical Learning” and “Contextual Adaptation”, as the US Defence Advanced Research Projects Agency defined them. But, as a universally agreed AI definition can hardly be identified, due to its multiple meanings and significances, the professors within the NPS consortium, the intelligence department, have come up with their own approach of what AI classification means, considering its learning power of calculating machines (getting a new action capacity). In other words, a “machine” is stronger than other if it can achieve a task that the “machine” it is compared with cannot.

In the AI learning process, there were identified the following levels: Level 0-Basic Automation; Level 1- Rule based systems; Level 2-Supervised Learning; Level 3-Unsupervised Learning; Level 4-Human-machine teaming; and Level 5-Apirational machines.

 Of all of these, the human-machine teaming is the level that maximizes, for now, the effectiveness and capabilities of the entire system. This is about planning and establishing some “intelligent” interfaces, which are allowing people and machines to team up and do their best, coming up with a better result than they could have individually accomplished. Because nothing we do today is purely manually or automatically, but in between, the human-machine teaming is the perfect balance between what the operator must do and what the machine can do better.  

Basically, human being should be responsible and effective when making decisions, in having and expressing preferences, being able to work in ambiguous situations, getting adapted to context, meanwhile machines should be responsible with making calculations, comparisons and applying logical judgments, with managing large sets of data, quickly, without getting bored, tired or starting to complain about a certain task, even if it was made billions of times, in maximum certainty conditions and lack of context.

There are no scientific proofs, in the military field, to claim that war is in human’s nature, that people were always included in the decision-making process, in the initiation, planning, training, organization, execution and, then, the evaluation of each military action, and the concern to permanently place the human in the middle of this process is still a currentness matter. Although some are trying to take advantage of using AI in the military field, we still want the human judgement to belong to commanders, so that they can be the ones to make the decision in terms of using machines or not, this way, executing the command-control authority.

Development programs of robot-tanks at a global level

In order to get and upkeep competitive advantages in the military competition, the great powers are competing in researching, developing, testing and endowing their military forces with new armament systems integrating AI, to allow operators to focus on the most important tasks, without wasting their energy on simple and repetitive functions, like driving a vehicle or charging the proper munition for a gun. Among these weaponry systems there is also the so-called tank robots, autonomous or semi-autonomous, manned or unmanned armored or on caterpillar vehicles, some of them dedicated to replace the main fight tanks, the track-borne canon or even the grenade launchers and missile platform on self-propelled mobile platforms.  

Capable of resisting to any high-intensity tough fight conditions and attack even the strongest objectives, the classic tank offers an impressive set of capacities on the modern battlefield. Being the main force to take down the army forces, it is able to execute missions in any weather conditions or tainted environments (nuclear, bacteriological or chemical), being really effective when it is using its fire power, mobility, protection and psychological shock action against the enemy.

In order to be effectively exploited, this classic fight method needs, however, many complex maintenance works, well-prepared operators, qualified machinists, as well as daily supply with: fuels, oils and lubes. Furthermore, the tank is more vulnerable to guns’ effect used by other wheeled or caterpillar armored platforms, fight helicopters, mines, antitank grenade and missiles launchers, antitank tunes and close support aviation, but also to infantry tanks, especially in urban warfare.

To overcome tanks’ limits, the researchers and experts in armored units fight are continuing to come up with creative solutions and create new autonomous or semi-autonomous tank robots, which are incorporating AI that will improve the fight power with its quick, simultaneous and synchronized action and implementation of some of its elements, such as: conduction, strike power, maneuver, protection, intelligence and logistic and fight support. These are all going to make more effective the breakaway operation of enemy’s lines, minimizing the risks of its own force (military men) and the mission. Projects by which such robot-tanks are able to simultaneously control small autonomous or semi-autonomous armored vehicles groups, execute the maneuver in the tactical field and execute patrol, research, surveillance or transport missions are starting to become more and more feasible from a technological perspective.

Therefore, in Israel, the Carmel program, which is related to tank’s future, is already starting to get some attention as the national defence industry companies have come up with a concept for a new tank to have a team composed of only two people (instead of the classic 4 operators team, commander, gunner, loader and the driver), teamed with multiple AI systems helping the team to identify the targets and to make decisions, almost completely automatized and with an improved visibility.

Technological solutions like the Iron Vision helmets, working with cameras and sensors mounted outside the tank, or cockpit’s transparent design to offer soldiers a 360 degrees view, augmented reality systems along the crew to quickly see the targets, their own forces or the disposal of important objectives, AI autonomous systems to support the missions, for mission’s autonomous planning, tank driving and functioning along with all vehicle’s armament systems, autonomous drone platforms operated with systems similar to video games consoles, cyber-defence systems and active camouflage, blue tracker systems, all dedicated to identify the targets, select the weapons, the autonomous move in different types of field or to help in the decision-making process are made by Israeli companies which are competing to get contracts for such military products for the armed forces.

Also, the Russian researchers and developers are working for quite some time on small wheeled and caterpillar robot-tanks, with machine guns and even missiles, on the conversion of armored carriers for troops in unmanned armored fight vehicles, equipped with turrets or remote-controlled missile or on the creation of robots with platforms and armament systems dedicated to team operation and managed from a command-control vehicle or by an operator who is 1km away. A joint tactical management network “Skynet” allows the connection of 4 robot-tanks, whether in a column or diffused on a 6 km distance.

The Americans think that, practically, all fight vehicles which are now being developed will be made with different new autonomy levels and will be able to use machine learning technologies, to process the key fight details, organize them independently and, eventually, send information to an operator who is executing the command and control. AI allows computers to immediately access complex data bases with millions of information to make real-time data analyses, before sending the military commanders the information integrated from multiples sources and analyzed so that to ease and hasten their decision-making process.

The American army forces have launched a new effort, called Project Quarterback, to accelerate armored fight units and subunits through the synchronization of battlefield data with AI. It is about an AI assistant, which can monitor the battlefield, taking all relevant data from drones, radars, ground robots, satellites, cameras mounted on soldiers’ glasses and which will be able to elaborate the best strategy to get the enemy out of the fight with their available weapons.  

The future plans of the army are mostly relying not only on AI, but also on more and more intelligent ground robots. Currently, only one operator of the American army can control two ground robots. However, they want get for one person to control 12 robots. The fight ground robots will have to be able to understand the environment they are in, calling on objects around them and to make decisions with a minimum human monitoring, as the network availability with bandwidth is rarely provided.

Developing such semi-autonomous armored platform is a natural foregoing phase towards the production of fight vehicles to be able to shoot, move and communicate autonomously. But these are all including some challenges. The data volume that can be used on the battlefield is quickly enlarging and needs, for example, a lot more time to be synchronized.

Furthermore, as the Russian have noticed, after testing the robot-tanks in fight, in the battlefield in Syria, there was a series of shortcomings which made them conclude that the robot-tanks will need more than 10 years of development until being able to effectively execute complex missions. We are talking about shortcomings in identifying and committing the enemy targets, after limiting the electro-optic devices and sensors of the lack of stabilization, the weak feasibility of the propulsion and suspension elements, which need, anyway, frequent and complex maintenance intervention, the ineffectiveness in the urban environment or rough terrain of the remote-control system of the robot-tanks, and the possibility for the operation executing the command and control to be exposed. These shortcomings are even more increased when, in battlefields like Syria, there is an extraordinary electromagnetic activity, determined by communication signals and connections between drones, large-scale jamming, espionage and other types of electronic war.

Instead of conclusions

Autonomous or semi-autonomous robot-tank are complex systems which, in order to be effectively exploited, need well-trained and well-instructed operators and the commanders and operators will have to be more and more confident in machines and software, producing results through processes developed through algorithms, created by human programmers, exploiting the human-machine work potential.

As I have found out during the NPS course’s speeches, although AI researches dreamt, for years, to create machines to work intelligently, it was always a different objective and, until now, the success was only seen in vehicles which work perfectly.

When it comes to comparing the human intelligence- AI development rhythm, in terms of accomplishing certain tasks, we have more and more proofs around us claiming that AI overcame the human intelligence. However, no one can guarantee that AI will ever reach the superintelligence level or the so-called “Technological Singularity[1]” level, developing both intelligence and consciousness.

Which will be these vehicles’ concerns in the future? What will be their connection with the human race? These are just some of the questions and concerns constantly emerging within the discussions on aspirations, but also concerns, related to superintelligence machines. Futurists, researchers and inventors as Ray Kurweil are claiming that the continuous exponential increase of calculus power will, by 2030 or 2040, lead to machines to have the complexity of a brain in a brain’s physical space.