Over the past few years – and particularly in the on-going war in Ukraine – we have seen the rise in use of what has become known as ‘kamikaze’ or ‘suicide’ drones. Two new excellent reports have just been released which examine these systems. ‘One-Way Attack Drones: Loitering Munitions of the Past and Present’ written by Dan Gettinger, formerly of the Bard Drone Center and ‘Loitering Munitions and Unpredictability’ by Ingvild Bode & Tom Watts, examine between them the history, current use, and growing concern about the increasing autonomy of such systems.
A drone by any other name…?
Firstly, to address the elephant in the room: are these systems ‘drones’?
Naming has always been a keenly fought aspect of the debate about drones, with sometimes bitter conflict over whether such platforms should be named ‘unmanned aerial vehicles’ (UAVS), ‘remotely piloted air systems’ (RPAS) or ‘drones’. ‘Drones’ has been the term that has stuck, particularly in mainstream media, but is regularly used interchangeably with UAV (with ‘unmanned’ being replaced in recent years by ‘uncrewed’ for obvious reasons). While many in the military now accept the term ‘drone’ and are happy to use it depending on the audience, some continue to insist that it belittles both capabilities of the system and those who operate them.
Whichever term is used, a further aspect of the naming debate is that an increasing number and type of military aerial systems are being labelled as ‘drones’. While all these systems have significant characteristics in common (aerial systems, unoccupied, used for surveillance/intelligence gathering and/or attack), they can also be very different in terms of size and range; can carry out very different missions; have different effects; and raise different legal and ethical issues.
One type of such system is the so-called ‘suicide’ or ‘kamikaze’ drone – perhaps better labelled ‘one way attack’ drone. There are several different categories of this type of drone, and while they are are used to carry out remote lethal attacks and therefore have significant aspects in common with the much larger Reaper or Bayraktar drones, they are significantly different in that they are not designed to be re-used, but rather are expendable as the warhead is part of the structure of the system which is destroyed in use. Importantly, while loitering munitions are a sub-set of ‘one way attack’ drones, not all one-way attack drones are loitering munitions.
Dan Gettinger’s report ‘One-Way Attack Drones: Loitering Munitions of the Past and Present’ helpfully sets out a history of the development of these systems and identifies three sub categories: anti-radar systems, portable or ‘backpackable’ systems and Iranian systems. He has compiled a helpful dataset of over 200 such systems (although not all are currently in operation). All of these, he argues, can be traced back to “the transition from the era of jet-powered target drones to that of remotely piloted vehicles.”
“The origins of today’s OWA drones are intimately related to those of the modern drone. In the 1970s and 1980s, the advent of new fabrication materials and lightweight sensors facilitated the production of ever-lighter airframes, hastening the end of a brief era dominated by expensive, jet-powered drones. The emergence of this new class of uncrewed aircraft eventually led to the adoption of drones for a multitude of lethal and non-lethal military missions, including for targeting adversary air defense radar sites. In time, the military applications of OWA drones broadened beyond that of the suppression of enemy air defenses. In the early 2000s, the emergence of the Switchblade provided the infantry with its own portable loitering munition, while the spread of Iranian drones enabled a host of non-state and, increasingly, state actors to use low-cost drones to conduct long-range precision strikes.”
Added to this, of course, can be the improvised arming of small, commercially available off-the-shelf drones which have been used by non-state groups in the Middle East and more recently, by Ukrainians to launch attacks against Russian forces.
The whole of Dan Gettinger’s report is well worth reading, but importantly he sets out a history of the development of such systems, tracing them back to efforts to US efforts to counter defensive radar systems in order to enable aircraft and missile attacks. “The growing sophistication of radar-controlled anti-aircraft weapons….. convinced the US and its allies to seek new technologies to curb their effectiveness.” By the mid-1970s, the Pentagon had a plan to develop ‘harassment drones’ which
“envisaged that low-cost drones equipped with a radar-homing seeker and a shaped charge would orbit over a pre-programmed area until a radar site began emitting electromagnetic radiation, whereupon it would dive into the site and detonate its charge. If the radar site ceased transmitting, the drone would continue loitering over the prescribed area to discourage the radar site’s operators from activating again.”
Throughout the remainder of the 1970s and 1980s various models and proposals were tested and evaluated by the US – including an important collaboration with West Germany to develop a system called Locust – but enthusiasm and funding for drones waned. By the early 1980s West Germany was working with Israel on anti-radar drones. As Gettinger writes “precisely when Israel first stared developing or acquiring expendable anti-radar drones is somewhat unclear.” However, by the mid-1980’s two very similar anti-radar drones with similar characteristics emerged from the relationship between Germany and Israel; the German DAR and the Israeli Harpy. While the DAR has not survived, today, Israeli company IAI promotes four different ‘loitering drones’ derived from the original Harpy: Harpy (New Generation), Mini-Harpy, Harop and Mini-Harop.
Backpack / vehicle launched attack drones
Separate from anti-radar drones are smaller one-way drones, either launched from vehicles or ‘backpack drones’ which were developed in order to provide soldiers on the ground with short-range aerial reconnaissance and/or attack capabilities without the need for expensive air support from the air force.
These small drones fly to the area directed by the operator who can view video feed from the drone and target in the attack. These drones too can ‘loiter’ above a target area from a short period of time and are therefore referred to as ‘loitering munitions’ The most well-known of these systems is the US Switchblade which has a range of around 10 km and can fly for about 15 minutes (although it should be noted that the US officially designates Switchblade as a missile system). Alongside these are larger drones that can be launched from a vehicle that are intended to target armoured vehicles and groups of personnel.
Iranian ‘suicide’ drones
Gettinger includes a third strand of development in his overview of one way attack drones: those developed by Iran. Interestingly, he notes that Iran’s first experience of UAVs was the import, during the 1970’s, of US target drones prior to the 1979 revolution. During the early 1980s Iran began building a number of UAVs for reconnaissance during the Iran-Iraq war, with Iran’s the first armed drone – the Ababil – coming after the end of the war. Variations of the Ababil attack drone have been used by a number of armed groups in the Middle East. Whether Iran has provided these systems to non-state groups directly, or provided basic plans to enable the development of local variants is disputed. In 2019, Houthi rebels launched a drone attacks on several Yemeni army bases killing a number of soldiers including the Chief of Staff. A September 2019 attack on Saudi oil processing facilities was claimed by Houthis, with US investigators assessing there were ‘similarities’ between the drones used and Iranian one-way drones.
A number of attacks on ships in the region have also been attributed to the newer Iranian Shahed family of one-way attack drones. These systems are not guided in by an operator but rather use basic satellite guidance and inertial measurement to hit a predetermined target. In late 2022, Iran agreed to export Shahed 131 and Shahed 136 drones to Russia, with numerous accounts of them having been used by Russian forces.
Last year, Drone Wars UK published a short piece looking at how the increasing use of loitering munitions in Ukraine raised concerns about the drift to autonomous weapons. Dr Ingvild Bode (The Center for War Studies, University of Southern Denmark) and Dr Tom Watt (Royal Holloway Centre for International Security) have written a new study ‘Loitering Munitions and Unpredictability: Autonomy in Weapon Systems and Challenges to Human Control‘ which dramatically expands and deepens the focus on this issue, examining the use of loitering munitions (both anti-radar and smaller systems) in relation to growing concerns about increasing autonomy in weapon systems.
Using a dataset of 24 specific systems and analysis of how loitering munitions have been used in three recent armed conflict (Libya, Nagorno-Karabakh and Ukraine) they argue that
“(T)he global practices of acquiring and operating loitering munitions clearly highlight the trend towards increasing autonomy in the targeting functions of weapon systems and how this affects human control over the use of force. We argue that the integration of automated or autonomous technologies in loitering munitions has created practical challenges and precedents regarding the quality and form of human control exercised over specific targeting decisions… This potentially makes human control over specific targeting decisions more nominal than meaningful. It also raises questions related to compliance with various legal and ethical norms.”
In particular they point to five specific aspects of how autonomy/AI can decrease “the quality and form of human control over targeting decisions” and relate these to the use of loitering munitions.
- Unpredictability: Loitering munitions are designed to loiter over a broad area, using sensors to search for potential targets before then attacking. This creates uncertainty and unpredictability over where and when lethal force is applied and erodes human control.
- Unpredictability in populated, urban areas: Some loitering munitions are now being specifically designed and marketed to be used in urban areas, with manufacturers claiming they can be highly efficient, precise and reduce civilian casualties. As Bode and Watts make clear, such claims overlook the increased levels of unpredictability in populated urban areas, with such complex environments being incredibly difficult – if not impossible – to observe reliably via remote sensors.
Loss of moral agency: While most current loitering munitions are designed with ‘human in the loop’ – that is, with an operator assessing via video feed or other sensor data the potential target to attack – it is not hard to envisage this aspect also being automated. As Bode and Watt write: “Given the challenges researchers face in verifying how loitering munitions with autonomous functions are operated in battlefield contexts, it would be difficult to know whether this threshold has been crossed.”
- Potential wide area effects: While loitering munitions are promoted as being precision weapons, a wide range of objects are potential targets if programmed into the system. While, as with larger and longer ranger drones, it is argued that the loiter time gives increased capacity to distinguish between civilians and combatants, interpretation of what is happening on the ground via remote sensors introduces real and concrete uncertainty.
- Cognitive overload: While in theory the use of autonomy and AI is in part aimed at relieving the operator from overburden, as Bode and Watt argue:
“Oftentimes, automated and autonomous technologies can make the task of human operators simultaneously minimal and more complex because it requires operators to have a functional understanding of both the anticipated outcomes of using systems integrated with autonomous functions and what the operational limits of these systems may be.”
This is even more of an issue when it comes to swarming technology. The authors quote WB Group, for example, who claim that a single operator can “effectively control” up to ten Warmate systems “simultaneously for co-ordinated attack.”
Bode and Watt’s study offers a number of recommendations in relation to controlling the development and use of autonomous weapons arguing that their investigation in to how loitering munitions work, as well as how they have already been used in a number of armed conflicts, points towards a future where the quality and form of human control over such weapons decreases.
The rising use of ‘one-way attack’ drones raises a number of serious global security challenges. Like their larger cousins, these often smaller systems make it much easier to engage in the use of lethal force; enable force to be used remotely thereby lowering the threat to operators but increasing risk to civilians; and also enable attacks to be undertaken covertly making accountability almost impossible. Already their use in urban areas of Ukraine has caused serious civilian casualties, while elsewhere they have enabled dramatic and damaging ‘spectacular’ type attacks. Added to these concerns must be the increasing autonomy of these systems which bring serious and challenging risks.
Both these studies greatly enhance public understanding of the use of these type of drones/loitering munitions and are well worth reading in full.