[Interview] From cheap drones to AI targeting: Ukraine’s robots still answer to human commanders

First published in Respekt. On the front line, it is not just a single best and most accurate weapon that decides, mass also matters, explained security expert Ondřej Rosendorf from the University of Hamburg. While drones have already rewritten strategy not only in the Russian‑Ukrainian conflict, Ukraine is now boasting a new weapon. Battalions of ground robots – tracked and wheeled machines that deliver supplies, transport ammunition, evacuate the wounded, lay mines and, increasingly, are said to be able to hold captured territory as well. Does this mean a further development in the ways wars are fought? For many people, the image of a war robot evokes the film Terminator. We are probably not quite there yet. So what do today’s frontline robots look like? [Rosendorf] They are usually unmanned systems, such as drones, which are either remotely controlled by an operator or incorporate automated or autonomous functions. When we talk about fighting robots, we mean precisely these autonomous weapons. The question, however, is what exactly they are. According to the Red Cross definition, these are weapon systems which, once activated, are capable of selecting, tracking, and neutralising a target on their own, without further human involvement. But this definition is problematic, because it also covers a number of systems that armed forces around the world already routinely operate today and that have existed for several decades. These include, for example, Shahed kamikaze drones used by Russia and Iran. Potentially, a number of cruise missiles would also fall into this category. This is exactly why experts point out that autonomous weapons represent a step further: this is a weapon that is capable of choosing by itself whom to attack, who is a legitimate target. At present, we have no evidence that anyone is using such a type of weapon. There are systems that incorporate autonomy at different stages or in different functions – but it is never the case that we have terminators walking around the battlefield and deciding on their own whom they will shoot. Ondřej Rosendorf is a researcher at the Institute for Peace Research and Security Policy at the University of Hamburg and at the Faculty of Social Sciences of Charles University. He also cooperates with the Peace Research Center Prague. His research focuses on preventive arms control and military innovation, autonomous weapon systems and international cooperation. Where does a human still decide and what does the machine already do by itself? For example, in FPV [first-person view] drones or kamikaze drones used by the Ukrainian army, almost the entire process is under human supervision. A drone may have automated or autonomous functions such as navigation. It is capable of getting from point A to point B. And there is something called last‑mile autonomy: just before the drone reaches its target, the operator switches on an autonomous targeting process and, over the last few metres or even kilometres, the drone finds the target by itself. The key point, however, is that the final decision is always made by a human, the operator. So, it does not actually choose the target by itself … No. The target is always defined in advance either by the operator, or, in the case of air‑defence systems, by pre‑defined parameters – that is, what constitutes a target they are supposed to hit. This means a projectile flying at a certain speed at a certain altitude. Then everything runs automatically, without human approval for each individual engagement; but it still holds that the operator can always switch off or intervene in the system. Ukraine is also using ground robots. At first mainly for resupply, because they can reach even mined areas, but it now claims that these robots are already able to hold defensive positions, or even capture enemy soldiers. The country wants to produce up to 50,000 of them by the end of the year. Will this be a new fundamental change and revolution on the battlefield, as we saw in the case of drones? It will not – and the main reason is quite simple. For ground robots, overcoming terrain is an enormous problem. Yes, Ukraine has a programme to manufacture ground robots and it is in its interest to boast about everything they can do. One can say this is part of wartime propaganda. Russia also has a programme to manufacture robots, but for comparison: both countries are producing millions of drones. So the difference in the role played by aerial unmanned systems compared with ground ones is enormous. A year or two ago, there were frequent reports that Ukrainians had very effectively deployed unmanned boats to destroy ships belonging to the Russian fleet in the Black Sea. Airborne and surface systems are relatively effective. For the ground ones, that issue of crossing terrain is simply a huge problem and they still represent a relatively small part of what matters to the Ukrainian army. However, they may prove useful in future, for example in the logistics just mentioned. Drones are not yet able to transport material or wounded soldiers from one place to another, and that is where I see potential for further innovation in ground robots. But it is still marginal. Ukraine is openly boasting about its ground robots. Do we know what the Russian ones can do in comparison? They are on the same level. The Russians have more resources than the Ukrainians, but in the end this does not give them a greater advantage. If ground robots today have any significant strategic role, it is primarily in demining. That is the main reason to develop them. The use of robotic ground platforms in combat is still far from reality. The fact that a drone or a robot carries a weapon from which someone can fire remotely is certainly interesting, but it is definitely not the main innovation on the battlefield. So what is the main innovation today?