7

Dull, Dirty, and Dangerous

at around noon on April 25, 2015, the buildings began to shake in Kathmandu. A magnitude 7.8 earthquake had struck the Gorkha region of Nepal, approximately 48 miles northwest of the city.¹ Within an hour, two additional aftershocks, measuring 6.6 and 6.7 on the Richter scale, struck the region, adding to the destruction and panic that followed. The earthquake was felt as far away as India and China and was the largest recorded in eighty years.² The worst affected areas were in Kathmandu and the surrounding regions, but the damage was hardly confined there. The United Nations estimated that the earthquakes affected eight million people, nearly a quarter of Nepal’s population.³ The death toll was 9,000, but could have been significantly higher if the earthquake had not occurred at lunchtime, when many people were away from their homes. A further 16,800 people suffered injuries and 2.8 million were displaced by the earthquake across the country. Perhaps more devastating was the fact that 600,000 homes were destroyed that day, with another 300,000 partially damaged.⁴ More than 90% of the homes near the epicenter were entirely flattened.⁵ Approximately 20% of Kathmandu’s most famous landmarks—including centuries-old temples of cultural and religious significance—were fully destroyed, with nearly all of the rest partially damaged.⁶ The earthquakes also caused a massive avalanche on Mount Everest, killing at least nineteen mountaineers and injuring sixty-one, while sending others fleeing for their lives.⁷ One experienced expedition leader later described a devasted base camp as looking like, “it had been flattened by a bomb.”⁸

In the affected regions, the problems seemed only to multiply. Thousands of those whose homes had been destroyed moved into the streets and lived in tent encampments. The days following the earthquake remained nerve-wracking for many displaced persons who ventured to their former homes to reclaim what was left behind. But no one was sure which structures were stable and which were only minutes away from collapsing. The misery was compounded by a pounding rain, which flooded the tent encampments and made the delivery of relief even more difficult. Aid workers who rushed to the scene recognized that the concentration of people in a wet, dirty, and rubble-strewn environment increased the risks of poor sanitation and disease.⁹ But the obstacles in the environment were considerable. The sprawl of tent cities holding displaced people and debris from collapsed buildings rendered some roads inside the city impassable. Outside Kathmandu, rural regions badly affected by the earthquake were cut off by landslides over key roads, while bad weather hampered helicopter access by relief agencies.¹⁰ Dozens of smaller aftershocks also afflicted the region and terrified the displaced residents, culminating in another magnitude 7.3 earthquake that struck the region in mid-May 2015.¹¹ That aftershock killed at least sixty-five people and injured another 1,900 more.¹² The urgency of the relief effort was underscored by the imminent arrival of monsoon season, which promised even more flooding and landslides.

In the immediate aftermath of the earthquake, Prime Minister Sushil Koirala declared that the Nepalese government would be on a “war footing” while it dealt with the relief efforts.¹³ But its actual record was significantly less vigorous. The government was overwhelmed by the difficulties of operating in such an environment and lacked enough manned helicopters for effective delivery of relief.¹⁴ Many of the displaced began to complain that they were being neglected by their government. The United Nations quickly established a flash appeal for Nepal earthquake relief, valued at an estimated $415 million, though the aid ultimately fell short of that total.¹⁵ A number of countries—including India, Bangladesh, China, the United Kingdom, and the United States—provided funds, food, and medicine and even military assets to relieve suffering and get the country back on its feet. The Indian military also played a crucial role in airlifting food and medicine into the country and coordinating the relief process. But the relief effort remained uncoordinated, with multiple governments, international relief agencies, and NGOs sweeping into the country, each with their own priorities. Even the airports were clogged with traffic from incoming relief planes and tourists fleeing the country.¹⁶

Among the people who arrived for the relief effort were a new class of “digital humanitarians” committed to finding ways to harness modern technology to alleviate the suffering of others in natural disasters.¹⁷ Some of the technology that they proposed using was fairly basic and widely accessible to people in developing countries. For example, almost everyone in Nepal had a mobile phone, and texts from those phones could identify where trapped people were, which roads were blocked by landslides, and where relief was needed most. If the affected population was connected to the Internet, even more opportunities for relief efforts based on what became known as “big data” became possible. For example, social media such as Twitter could be used to glean information about the damage and, if properly geo-located, create a more textured picture of the regional impact of the disaster. But the digital humanitarians were also committed to using drones to help identify which buildings had been damaged and which roads were blocked and holding up relief efforts. A number of international agencies and NGOs soon arrived in Nepal equipped with small but capable drones to aid in relief efforts. These drones were used to identify victims of the earthquake for search and rescue missions, to evaluate ruined heritage sites, and also to take images of damaged buildings so that they could be assessed to see how likely they were to collapse.¹⁸ Aerial imagery could be used to create 3-D models of buildings and provide a more accurate picture of which buildings were likely to collapse given the structural damage they had suffered in the earthquake.¹⁹ In this new “digital humanitarian” model, data could be aggregated from multiple drones and other sources and collectively mapped using a software application such as Micromapper or OpenStreetMap. This would allow effective “crowdsourcing” from multiple UAVs and rapid delivery of the data to aid agencies and other relief organizations.

The advantages of this approach were clear. Digital humanitarians were able to provide better, more detailed images of crumbling buildings and blocked roads than satellites, and this additional level of detail saved time and effort for relief organizations operating in a difficult, time-sensitive environment. But these operations in Nepal were hardly an unqualified success. Nepal had effectively no regulations on drones at the time of the disaster, and as a result many operators were confused about the rules for flying them. Were drones nothing more than hobbyist devices that could be used freely, or did they require the approval of the civil aviation authorities? While most organizations behaved responsibly, some flew drones without checking with local leaders or governments. Others flew too close to military installations and historic sites, while others inadvertently buzzed over and alarmed groups of traumatized people. The Nepalese government, already overwhelmed with the scale of the disaster, found some of these “disaster drones” an added worry rather than a relief. Yadav Koirala, an official with the Home Ministry of Nepal, said later that “drones were an added burden to us. There was no restriction on them . . . They flew everywhere but offered no feedback to the government . . . It was difficult to control them.”²⁰Some Nepalese officials worried that broadcasting images of the destruction wrought by the earthquake would damage the country’s valuable tourism industry.²¹ Other agencies worried about a collision between a drone and a manned aircraft around Kathmandu’s airport. The Nepalese military and intelligence establishment became alarmed about the prospect of spies using the humanitarian drones as a cover, a fear particularly aggravated when a Chinese national was caught filming army headquarters with a drone.²² In the middle of the humanitarian effort, Nepal banned drones from flying for more than fifteen minutes and issued no-fly zones for security buildings, houses, and religious and cultural landmarks. The regulations also stipulated that drones could be flown no higher than 100 meters and no more than 300 meters from the pilot.²³

The Nepalese earthquake had become a testing ground for digital humanitarians to use drones for good purposes, but the effort had run aground because of ambition, lack of coordination, and carelessness. Yet it also raised a series of important questions about the nature of humanitarian work in the drone age. The digital humanitarianism movement seeks to take advantage of the capabilities of drones to do the kind of work—often described as dull, dirty, or dangerous—that others are incapable of or unwilling to do. Their efforts are designed to show that drones are empowering international and local actors in humanitarian disasters in ways never before imagined. Some advocates point to the spread of drones as democratizing data collection—in other words, stripping it from the hands of often selfish or parochial governments—and making the world “a better, safer place.”²⁴ Yet despite good intentions, some elements of the digital humanitarian movement are suffused with a technological utopianism that conflates what can be done with what should be done. Drones have unleashed new capabilities for those who wish to help, but they have also complicated the humanitarian space by introducing new actors whose incentives may not align with the local government or population. As the Nepal earthquake disaster revealed, there is a risk that the needs of the local population would be swept aside by organizations more eager to demonstrate the proof of concept for humanitarian drones than to realistically address the needs of those suffering. All of those advocating for drones as a force for good—digital humanitarians, activists, and even UN peacekeepers—need to be careful that they do not lose sight of their goals or become more risk-taking as a result of access to the technology. For “disaster drones” to be successful, they will need to prove that new technology like drones are necessary for achieving humanitarian objectives, rather than just an opportunity to test the capabilities of new technology amid human suffering.

Disaster Drones

The Nepal earthquake was not the first time that disaster drones were deployed in a natural disaster or emergency. As with most things related to drones, this began with the military. Immediately following a massive earthquake in Haiti in 2010, the United States provided overflights by Predator and eventually Global Hawk drones and released satellite imagery which aided the relief efforts.²⁵ While this assistance was useful, the government of Haiti remained uneasy about the unfettered use of small drones over the disaster zone. Its initial reaction was to ban small UAVs for fear they would collide with civilian air traffic moving through the airspace for the relief effort. Later, in 2012, the International Organization for Migration (IOM) received approval to use smaller drones to assess the damage done to Haiti after its deadly earthquake. In cooperation with the United Nations, the IOM used drones to identify debris hazards, new construction, water drainage capabilities, and temporary housing in Port-au-Prince and the surrounding region.²⁶ The IOM used drones to assess destroyed structures, to conduct surveys of public buildings, and to monitor internally displaced people (IDPs) and camps.²⁷ Images supplied by drones constituted an improvement over satellite imagery, which by some estimates tended to underestimate the scale of the damage in Port-au-Prince.²⁸ The results, according to the UN project leader, were “incredibly satisfying” and also notable because the Haitian displaced population was curious rather than hostile when seeing drones float by overhead.²⁹

In the Philippines, drones were also used to assess the damage caused by Typhoon Haiyan in 2013.³⁰ In this instance, one of the biggest problems was that the satellite imagery was outdated and some communities were relying on hand-drawn maps.³¹ To address this situation, a number of organizations sponsored projects and volunteered their time and resources to improve the quality of images available on the ground. Medair and Catholic Relief Services partnered to use drones to collect aerial imagery in the reconstruction efforts after the disaster.³² A comparatively new organization, the Humanitarian OpenStreetMap Team (HOT), used satellite imagery donated by a range of actors such as Yahoo!, Microsoft, Digital Globe, and the US State Department to create a picture of where the hardest hit regions were and which roads were considered impassable. They were also helped by the fact that the cost of satellite imagery had plummeted over the last decade and as a result was now used by a greater number of organizations and not just governments.³³ By one estimate, over 1,500 volunteers in eighty-two countries participated in the effort to collect and map recent satellite imagery onto the terrain.³⁴ With this data as a base, they were able to add UAV imagery from local non-governmental organizations (NGOs) and create a detailed, street-by-street picture of the impact of the disaster. The chief obstacle that NGOs faced was a lack of coordination; many of these organizations did not know each other and were not aware of each other’s activities, which in turn undermined the effectiveness of the collective relief effort.³⁵ Yet despite these limitations the response to Typhoon Haiyan suggested that crisis mapping by volunteers coordinated largely over the internet could pay dividends for relief efforts. One assessment of drone use in the response to Typhoon Haiyan concluded that drones were able to speed up relief efforts, eliminate wasted time and effort, and improve the accuracy of the delivery of food and medicine to those suffering.³⁶

For many in the digital humanitarian community, Haiti and the Philippines served as proofs of concept for the value of crisis mapping using drones. Similar crisis mapping projects were launched in the aftermath of Cyclone Pam in Vanuatu in 2015 and a series of earthquakes that rocked Ecuador in 2016.³⁷ There, drones were used to monitor the stability of roads and even assess the risk of landslides.³⁸n 2015, the World Bank funded a series of drones used by the NGO Drone Adventures to map flood-prone areas of Dar-es-Salaam in Tanzania in order to improve crisis response.³⁹ In 2015–2016, the European Union began to experiment with drones for search and rescue missions. Many of these projects were collaborative and drew technology and insight from private companies, NGOs, and universities, with funding from major corporate donors and international institutions like the World Bank. At the same time, networks of roboticists and development professionals emerged to coordinate the use of UAVs for humanitarian purposes. Among the most notable of these is the UAViators, the Humanitarian UAV Network, which seeks to disseminate regulations about the operating environment, best practices, and lessons learned from these missions. In 2017, this network boasted over 500 pilots in seventy countries and served as a hub where calls for humanitarian drone pilots could be made in the wake of a natural disaster.⁴⁰

The success of these disaster drones hinges on both technical and human requirements. At a minimum, it is critical to deploy the right kind of drone—a number of different types of drones have been deployed for these types of operations. Multirotor drones, such as the popular DJI Phantom quadcopter, are simple to launch and do not require much space for take-off; they can be bought cheaply and used easily in urban areas. Their chief limitation is a limited battery life; they can only remain in the air for a short time.⁴¹ Fixed-wing drones can carry a heavier load and fly for a longer period, but they need a launching strip, which can be difficult to find in dense urban or mountainous regions.⁴² One particularly promising fixed-wing model, the Sensefly eBee, has a longer range, high levels of automation, and can remain in the air for a longer flying time than other drones. It has been used in Haiti and Nepal.⁴³ But it is not as simple as throwing a drone into the sky. For effective crisis mapping, drones need to be equipped with cameras with powerful resolution to get the kind of detailed images that are needed to depict on-the-ground damage. In general, drones flying at high altitude often get lower levels of resolution for their images, but those flown at lower altitude need to have a higher resolution camera and can generally cover less ground. The resolution of the camera can vary depending on size and cost. Visible light or RGB (red green blue) cameras are often better for crisis mapping because their pictures are similar to satellite images and can be more easily interpreted by aid workers.⁴⁴ These drones must also have a gimbal to stabilize the camera during flight. Aside from cameras, drones can also carry infrared sensors which can detect vegetation and water.⁴⁵ For drones to be viable in humanitarian crises, drones typically need some professional-grade features, which raises the costs involved and rules out some of the popular off-the-shelf commercial models.

The decision by a government or NGO to use a drone is partly dependent on the alternatives available. In some circumstances crisis mapping could be more effectively or cheaply done with satellite images or even pictures taken from helicopters. The reduced price of satellite imagery on the commercial market, from vendors like Google Earth, gives a clear advantage to it over drones. This is particularly so if satellites or military-grade drones can cover more ground. For example, the US military’s Global Hawk drone can take wide area shots, akin to shining a set of floodlights on a sports stadium. By contrast, most disaster drones cover far less ground and are closer to shining a spotlight on that same sports stadium. One estimate suggests that on average disaster drones can cover approximately 3 sq. km in a day, or 10 sq. km a week, for data collection and processing.⁴⁶ However, one of the most important advantages disaster drones provide over satellites is that they can capture images below clouds. Some commercial satellites are not equipped with cameras that can peer beneath cloud cover and see the ground below. This is particularly crucial for disaster relief because major storms like hurricanes and cyclones come with dense cloud cover that can obstruct satellite imagery for days or even weeks.

For disaster drones, what matters most is the quality of images and their correspondence with the real world so that they highlight areas where relief efforts are needed. For this to work, images taken from drones need to be spatially accurate and related to real-world maps. Most disaster drones are equipped with lightweight GPS units that are capable of locating exactly where an image was taken. These images can be combined with widely available satellite images and geometrically corrected to become uniform so that they adhere to common geographic coordinates and “knit together” on a single map.⁴⁷ The creation of what is known as a georectified orthomosaic—essentially a distance-corrected map that corresponds to what is actually on the ground—sounds more complicated than it is. There are many free or cheap software packages that can help crisis mappers create these maps on their laptops. It is particularly important to make sure that normal two-dimensional pictures do not distort the distance between two points—such as towns or buildings—on a map. It is not hard to imagine how important this is for time-sensitive relief efforts that depend on traversing large distances accurately to deliver food, medicine, or other supplies to needy populations. Disaster drones can also produce other useful images. For example, some drones are equipped with software that allows them to create 3-D models based on a calculation of the volume of structures appearing in images, which are useful for assessing whether buildings are damaged or structurally unsound and should be evacuated. Thermal maps drawn from drones are also useful in detecting structural damage to roads or the source of groundwater discharges.⁴⁸

This imagery, once collected, must still be processed quickly and delivered to the relief agencies to be useful. This is one of the chief limitations of disaster drones. At present, in many disaster zones, the infrastructure and the drones themselves are not present on the ground and must be brought in by external parties. One estimate suggests that drones take 6.5 days to arrive to the scene of the disaster, well beyond the crucial seventy-two-hour window when most relief is needed.⁴⁹ To be most effective, drones need to be in place before the disaster or delivered rapidly immediately thereafter. This is often not the case in the disaster-prone parts of the developing world. Most of the drone labs employed by digital humanitarians are in major hubs of the developed world, specifically in North America, Europe, and parts of Asia. The teams of volunteers and the drones themselves must get to the disaster zone quickly, which is difficult if commercial air travel is limited or expensive. Even when the teams of volunteers and drones are in place, processing the imagery quickly enough to be useful presents a challenge. Large-scale data collection of thousands of images, or the collection of particularly high resolution images, can slow the processing time considerably. There may be days or even weeks between when the images are taken and when they are developed and returned to the disaster zone. This time delay is less important if the drone project is itself not urgent. For example, some drone humanitarian projects are proactive, mapping areas of a region that might be susceptible to flooding or landslides. In these cases, the delay is unfortunate but does not detract from the value of the project. But for immediate disaster relief, when lives are at stake and the risks multiply as the days pass, getting disaster drones to the right location and analyzing images and delivering them to relief workers is a time-sensitive proposition. In these circumstances, the arrival of disaster drones can sometimes be too little, too late.

Aware of this fact, some organizations have rushed into disaster zones with drone projects designed to shorten the time that people must wait for relief. This was the case in the Nepal earthquake, where frequent media coverage of the disaster drew a number of NGOs and private organizations to the country and led to drones flying unchecked everywhere. This rush to the disaster zone is in part a function of the technology itself: cheap and readily available drones empower small groups of individuals who might otherwise have no direct way to act in a crisis. Many of these teams are well meaning, but such a rush to a disaster zone provides a space for opportunists and those less interested in helping people than experimenting with the technology to arrive. Moreover, even the best and most skilled drone teams are not always coordinated, either with the government or with each other. In Nepal, this produced a strong backlash and made some people increasingly suspicious of drone humanitarianism. Patrick Meier, the founder of the Humanitarian UAV Network, called the rogue drone pilots “cowboys” and estimated that a lack of cultural sensitivity and respect for the local population set the humanitarian UAV movement back at least six months.⁵⁰ In response to the problems seen in Nepal, Meier and his colleagues generated a database of UAV regulations by country and produced a series of guidelines for drone pilots to ensure that their drones were used in a way that respected the preferences of the local population. This development of a code of conduct, as well as the training and skill-building associated with it, will go a long way toward ensuring that disaster drones are used more responsibly in the future. But at the most fundamental level these codes of conduct remain voluntary, and there remains a danger with drone humanitarianism that some organizations will ignore the guidelines and laws and run roughshod over governments that are otherwise distracted and incapable of enforcing their laws on UAV use.

This is particularly important because disaster drones are not used in a normal environment where their presence might be otherwise expected. In some parts of the developing world, drones are not commonly seen and their sudden appearance can alarm the population. This is particularly the case in environments where high-end technology like drones typically belongs to the military. Especially in countries in which the military is not widely trusted, drones flying overhead might be seen as an attempt to conduct surveillance on the population for unknown ends. When dealing with a traumatized population which has recently experienced a natural disaster, these concerns over the ultimate purpose of drones are magnified. The last thing that humanitarian UAV users would want to do is to alarm or even terrify a population which has been already coping with a natural disaster. One way that this might be remedied is to paint the drones a specific color to signal their humanitarian purpose and to distinguish them from military drones. But such a measure might not be enough to put some lingering unease about the technology to rest. While more than a majority of aid workers (60%) surveyed in a 2016 study had a broadly positive view of drones in humanitarian work, a minority raised concerns that drones might be perceived as “inhumane” or too distant from the population to be useful. Others expressed concerns that their use will contribute to the trend of delivering aid at a distance and removing needed aid workers from dangerous environments.⁵¹

Many of the advocates of humanitarian UAVs recognized that it is crucial that drones not be seen by the local population as a foreign object, alien or otherwise hostile, but instead see them as part of an effort that has a human face. One way that this is being addressed is by investing in partnerships with local NGOs and universities so that drone humanitarian projects come from the societies that they are helping. WeRobotics, an NGO based in Switzerland and the United States, builds partnerships with universities, governments, NGOs, and other community organizations in the developing world to identify places where drones and other robotics might be useful and to develop projects to meet local needs.⁵² The goal is to create “Flying Labs” where local organizations can develop hubs to train local people to use the technology and teach others. WeRobotics currently has projects underway in Nepal and Tanzania and plans for as many as fifty more.⁵³ In 2019, in a partnership with the Centers for Disease Control and Prevention (CDC), they were also engaging in training and capacity building for cargo drone use in Papua New Guinea.⁵⁴ By undertaking these projects, WeRobotics addresses some of the concerns about the sustainability of these drone projects and builds capacity and buy-in by the local population and its representatives. This is a crucial step: to turn people in the developing world from objects of the drone’s gaze into agents in charge of what the technology does next. Such local ownership and buy-in will address cultural and social issues surrounding drone use and make humanitarian relief through drones more effective and sustainable.

There is no doubt that disaster drones have empowered new actors to take images of crisis zones, but it is still unclear whether their use will produce a sustained and large scale benefit in humanitarian disasters. Most of the cases in which disaster drones have been used could be described as pilot projects where the intended scale of the impact was small. For this reason, there is a risk that drone humanitarianism may become an example of tokenism—that is, when organizations invest in a project that does only a small amount of good and is directed more at proving the value of the technology than alleviating human suffering. Even if this does not happen, it is unclear that disaster drones can be scaled upward in a cost-efficient way to produce the kind of sustained or systematic change in disaster zones that their most optimistic advocates promise. It is also unclear whether all drone humanitarian projects will adopt the goals of sustainability and local buy-in that WeRobotics and others endorse. But the promise of disaster drones has only recently come into view. While there is no conclusive proof that providing more information and improved maps will vastly improve the response of governments, IOs, and NGOs to humanitarian disasters, there is more than enough evidence to support modest investment in drone projects that will complement existing capabilities (through satellites and other means of learning what is on the ground) and allow for a precise and carefully tailored response to the needs of those affected by disaster.

Search and Rescue

Another function of drones in humanitarian emergencies is search and rescue. The advantages of drones for this function are clear: they can cover large swaths of ground, operate beneath the cloud cover that typically obstructs satellites, and with the right cameras they can see through some obstacles (like tree cover) to find missing or endangered people. It is not hard to imagine the usefulness of drones in finding missing children, for example, or lost hikers in a vast national park. In humanitarian crises, the same technology could eventually be put to work in finding those stranded or endangered by floods and other disasters. The problem is largely one of access. Today, most NGOs and IOs operating with drone humanitarian teams do not yet have the capacity to conduct effective search and rescue. In part, this is due to timing: it takes too long, on average for a drone humanitarian team to arrive in a crisis zone for them to be useful in most search and rescue missions. In the aftermath of an earthquake or flood, the chance of survival for those lost and trapped declines precipitously as each day passes. In the little less than a week it takes for most drone humanitarians to arrive in a crisis zone, those in most need of help may have already been found or even have died. Seen from another light, the problem is also one of location. Some of the world’s worst disaster zones—places where earthquakes and flooding, for instance, are recurring—are located in the developing world where the infrastructure both for delivering relief and for supporting humanitarian drones is embryonic. If organizations like WeRobotics succeed in boosting local capacity for drone use in the disaster zones of the developing world, search and rescue will become more feasible than it is today.

At present, most of the sustained efforts for using drones in search and rescue have been pilot programs by police and civil defense forces in the developed world. In the United States, for example, drones have been purchased by a large number of police forces responsible for extensive rural areas like Texas and the Grand Canyon. In the United Kingdom, the Greater Manchester Fire and Rescue Service has flown Skyranger drones since July 2015 and have a dedicated Aerial Imagery Reconnaissance Unit (AIR Unit) to provide overhead imagery of incidents and their ensuing rescue operations.⁵⁵ In Europe, the European Emergency Number Association (EENA) signed a partnership with the company DJI to train first responders to use Phantom and Inspire drones as a first response following a disaster and to find missing people.⁵⁶ The European Commission has recently devoted €17.5 million to the Integrated Components for Assisted Rescue and Unmanned Search (ICARUS) program, which equips first responders with drones and other robotics to respond to emergencies.⁵⁷ The ICARUS program also promotes the development of drones and other tools and integrates them into the standard operating procedures of first responders across Europe.

These efforts at using drones for search and rescue are being tested by the refugee crisis that has engulfed southern Europe since 2012. Thousands of refugees, fleeing growing chaos across the Middle East and North Africa, arrive annually in Europe from Syria, Iraq, Libya, and Afghanistan.⁵⁸ In 2015, according to FRONTEX, the European Union border and coastguard agency responsible for monitoring refugee movements, 1.8 million refugees tried to enter EU member states, often via water routes across the Aegean or the Mediterranean Sea.⁵⁹ For many European governments, the influx of war refugees is potentially destabilizing, especially if the refugees are able to claim legal right to remain in their countries due to the threat of harm if they were returned. For this reason, the European Union has invested in interdiction of refugees at seas, often with patrol boats in the Mediterranean, which are designed to stop traffickers moving people into Europe.

Leaving the moral and legal issues aside, the process of interdiction at sea is challenging from a technological vantage point: patrol boats must find small vessels and dinghies that move under the cover of darkness toward entry points into Europe. Satellite images are ineffective because it takes too long to produce and update them—by which time vessels detected have moved on and are lost. From a search and rescue vantage point, timing is crucial. Some of the ships carrying refugees capsize in rough weather, leaving them drowning at sea and creating an immediate need for emergency response. The UN High Commission for Refugees (UNHCR) estimated that 5,000 refugees died crossing the Mediterranean Sea in 2016 alone, making it one of the deadliest years for refugees in history.⁶⁰ To deal with this crisis, the European Commission proposed a plan, worth €22 million, to set up an EU drone fleet equipped with video feeds, sensors, and chemical “sniffers.” By September 2018, FRONTEX was beginning to test drones in Greece, Italy, and Portugal to monitor the EU’s borders. The purpose of the EU drone fleet is not exclusively humanitarian; the “sniffers” are included to catch boats that are violating EU emissions standards.⁶¹ But the drones are also designed to find smaller, makeshift crafts during day and nighttime hours.⁶² The EU drones are equipped with thermal sensors, which make spotting refugees at night easier, and can pick up the heat signatures of refugees swept overboard at night and at risk of drowning.⁶³

Although FRONTEX remains the chief organization for managing the European Union’s approach to the refugee crisis, another agency—the European Maritime Safety Agency (EMSA), based in Lisbon, Portugal—has been tasked with arranging for drones for this purpose. EMSA has a long record in enforcing maritime safety and environmental regulations, but the humanitarian dimension of its new mission is relatively novel.⁶⁴ Initial efforts to turn this proposal for humanitarian search and rescue drones into reality has run up against a host of obstacles, not the least of which is developing coherent EU-wide regulations to integrate drones into the civilian airspace.⁶⁵ Across the European Union, there is a reluctance to embrace “good” drones given the negative connotations that the technology carries due to its use for military operations and targeted killings by the United States. But perhaps most importantly, the effort to use drones for humanitarian purposes has run up against questions of cost: large military drones such as the Reaper or Global Hawk are too large and expensive to be viable for these tasks, while small commercial drones are too limited in capacity to be useful. In effect, organizations like EMSA need a drone that has the endurance of some of the military drones, but with fewer capacities for holding payloads and a lower price tag.⁶⁶ EMSA is confident that it will be eventually able to find a supplier to provide these drones and service the ground stations needed to run them, thus allowing for cost-efficient rescue operations at sea, which will compliment rather than replace its existing operations.

The advantages of drones for this particular search and rescue mission are clear. Drones can cover a wide area, detect vessels and people that are hard to see with satellites and ordinary cameras, and communicate almost instantaneously with ground control stations. Drones can also approach human trafficking vessels while making very little noise, which is important because the sound of the approach of manned aircraft often makes the traffickers flee to avoid detection and capture at sea. Drones can also last longer at sea and fly for hours without the need to allow crew to rest. But the ambitions of those planning to use drones for search and rescue are not limited to detection by sight alone. Leendert Bal, director of search and rescue operations for EMSA, envisions a future in which drones may be able to detect mobile phone signatures or SIM cards of refugees at sea.⁶⁷ Sensors that were able to detect and locate mobile phone signatures would reveal where people are even in low visibility environments, such as rescue operations at night or in bad weather. Another longer-term objective would be for drones to drop life vests or even small rubber boats to help refugees whose boats have capsized.⁶⁸

While the European Union’s efforts at integrating drones into its search and rescue operations are embryonic, they are an important indicator for the new capabilities this technology provides to humanitarian actors and changes the actions they will consider as appropriate. For most actors, drones will be just another tool in the toolbox for conducting effective search and rescue operations. But in the case of the European Union, the technology is moving the boundaries of what is considered legal and appropriate as well. While these search and rescue drones in the Mediterranean have been cast by their defenders as the quintessential example of humanitarian drone use, they are used in the service of a refugee policy that is widely criticized as inhumane by organizations like Human Rights Watch and others.⁶⁹ The European Union’s efforts to deploy drones to maximize situational awareness could be read as an attempt to deny the refugees’ right to asylum on the continent and, in some cases, to forcibly return them to dangerous environments.⁷⁰ Although it is described as “search and rescue,” it is part and parcel of the European Union’s efforts to police its border regions against refugee claims. This illustrates that the technology is itself not value neutral, as it enables a policy which may have some legal basis but is considered ethically questionable. At a minimum, drone technology cannot sort those who might have a right to stay from those who have no such rights, as this is a legal question. Instead, the deployment of drones offers the misleading appearance of control over a refugee situation that is rapidly worsening with every year that passes.⁷¹While the camera on the drone itself yields precise images, it enables a humanitarian policy conducted at a remove from those who suffer the most.

Delivery Drones

The next stage in the evolution of the humanitarian drone will involve delivering parcels of food and medicine to populations in need. At present, most of the discussion of delivery drones has revolved around the plans of large companies like Amazon to deliver consumer packages to residences in the United States and Europe. At present, these plans are unlikely to revolutionize commerce in the developing world as some of their advocates claim. Delivery of ordinary lightweight packages by drones may be less cost-efficient than delivering products by ordinary truck in many densely populated areas. In part, it depends on the cost of drone deliveries. At present, the cost estimates are highly speculative ($1–6 per package), but unless the cost curve changes in favor of drones—and it might, as one analyst suggests Amazon can get the cost down to 0.88 cents per package—it will be hard to compete with trucks that can deliver hundreds of packages per day.⁷² This is even a bigger problem for heavier packages (over 5 pounds) which are more challenging and expensive to move. Drones will also need to be equipped with “see and avoid” technology and insured against accidents or theft in many residential environments, a factor which may alter the calculation of cost-efficiency for commercial drone delivery. There is also a thicket of safety regulations and laws that delivery drones must contend with in most locations in the developed world. Although start-up companies like Flirtey are convinced that there is potential in commercial drone delivery in the United States and other places, it will take regulatory change and considerable investment by major commercial players like Amazon or Walmart and a reduction in costs of drone deliveries to turn this dream into a reality.⁷³

At present, the greatest potential of delivery drones does not lie in getting packages or novelty goods to relatively well-off consumers in US cities. Rather, it lies in the efficient delivery of medicine, blood, and other essential goods to populations who lack effective road networks or for whom the delivery of goods by truck is cost inefficient. Consider, for example, how expensive it is to deliver an Amazon package to a customer in a city compared to a customer in a rural area. A truck might make dozens of package deliveries in the same few blocks in New York City, thus making the costs involved—for example, the labor of the driver and the costs of running the truck—worth it from the company’s perspective. But driving a single package to a customer a hundred miles or more from the delivery center is expensive, costing the company much more for the hours of labor of the driver and gas for the truck than they might get back in shipping charges. In these cases, there is a compelling cost-efficiency argument for using a drone rather than sending a driver out to spend several hours delivering one package.

Such remote populations are often in rural areas in the developing world, but they can also be located in geographically isolated parts of countries in the developed world. Delivering to remote populations is often called “last mile” delivery because it takes the package for the last section of a supply route. In parts of the developing world, this “last mile” delivery may mean the difference between life and death. There, many routine diseases become life-threatening either because vaccines and medicines are not available or slow to reach those in need, or because medical test results take so long to be delivered that treatment starts late, if at all. In these cases, delivery drones could be successful if they shorten the time periods between the emergence of symptoms, their diagnosis, and their treatment. By some estimates, one third of the world’s population lacks access to essential medicine, in part due to poor roads and the failure of “last mile” supply networks.⁷⁴ As a result, approximately 40% of vaccines supplied to people in the developing world expire before they are delivered and ready for use.⁷⁵ If the supply chain could be tightened by the use of drones, medicine and test results could be delivered rapidly and the gap between diagnosis and treatment would become smaller.

The technology behind mass-scale drone delivery is not quite ready, but it shows promise for specific kinds of payloads. Most delivery drones can hold only a few kilograms of weight and fly a relatively short distance, often less than 150 km.⁷⁶ To be effective, experts estimate that delivery drones should be sent out on short-range missions (typically 40 km or less) to help in areas with enough population density to make the trip worthwhile.⁷⁷ But one of the chief constraints facing such drone use is that the infrastructure for large-scale delivery drones simply does not exist at the commercial level. While the US military has made large deliveries of up to 2,700 kg with K-Max or Snowgoose large cargo drones, this movement of bulk cargo is not possible or cost-efficient for many NGOs and other organizations with less budgetary largesse.⁷⁸ Instead, most off-the-shelf drones are best used for high-value, low-weight products (blood samples, vaccines, tests) and some food parcels, water, mobile phones, and other lightweight, highly useful goods.⁷⁹ As a general rule, the heavier the payload, the harder it is to have a cost-efficient and reliable delivery to underserved parts of the developing world. While this might change in the future, adding additional payload to a delivery drone certainly increases the costs associated with its operation.

Other technological limitations make drone deliveries harder than simply flying and dropping the cargo. Many of the existing delivery drones are operated by rechargeable batteries, which must sustain the entire return flight or be designed so that they can be recharged at a waystation somewhere along the flight route. If the journey is too long, pilots must pre-determine a recharging station along the route and have enough people and infrastructure there to recharge and relaunch the drone. This means that even medium-range drone deliveries require a sophisticated operational plan, as well as plans for what could go wrong, before payload can be effectively moved. But perhaps the biggest issues that drone delivery advocates face is take-off and landing. Most fixed-wing drones require a runway for take-off and landing, which does not always exist in the locations that need the delivery. When these runways are available, fixed-wing drones are a preferred option because they can fly for longer and carry more payload than comparable quadcopters. When runways are not available on either end of a proposed delivery, pilots must be satisfied with quadcopters that are capable of vertical take-off and landing, but can carry fewer kilograms of payload and fly a shorter distance. In both cases, one of the key questions is whether the drone should drop its cargo in mid-flight or land to deliver it directly. The latter is safer and more reliable, but once on the ground drones can be stolen, hacked, or destroyed rather than returned safely. Dropping cargo from the sky is riskier and may miss the target but does not carry the same risk of deliberate or malicious disruption.

Efforts to use delivery drones in rural regions of the developing world have been underway for some time, though relatively few projects have moved beyond the proof of concept stage. A Silicon Valley start-up called Matternet has been deeply involved in some test cases using its M2 quadcopter drone, which has substantial battery life and intuitive handling. In 2014, Médecins Sans Frontières (MSF) worked with Matternet to experiment with the use of small, payload-bearing drones to transport the test results of those suspected of contracting tuberculosis in Papua New Guinea.⁸⁰ A similar effort was made by the World Health Organization and Matternet to use drones to connect hospitals in the urban and rural regions of Bhutan.⁸¹ In 2016, a UNICEF-Matternet joint project in Malawi delivered laboratory samples for early infant diagnosis of HIV faster than motorbikes over distances up to 10 km.⁸² In early 2016, a South Korean company called Angel Wing delivered medicine, needles, and vaccine vials to rural health clinics in Nepal.⁸³ Most of these projects were small in scale, testing the concept of operations rather than making a sustained impact, and with modest operational ranges. Yet the operations themselves pointed to greater potential in scaling up these projects for a much wider impact.

One way that this will happen is by investing in local partners who are able to assist in getting a project off the ground. In 2015, the United Nations Population Fund (UNFPA) worked with Drones for Development, an Amsterdam-based start-up, to use Dr. One, a hybrid drone designed to deliver contraceptives to rural clinics in Ghana. One advantage of this program was that it was done in partnership with Ghana’s health service. The Dr. One drone is designed to use lightweight material and open source electronics to allow local operators to use and maintain the drones at low cost. The project resulted in savings of about €3,000 compared to similar operations with motorbikes. There are still obstacles to scaling this project up to the national level—for example, a central mechanism needs to be established to coordinate and run the drone flights and the supply chain on both ends must be made more efficient for different types of payloads—but the Dr. One model showed that it could work.⁸⁴It also showed that some drone models could be made in a flexible way to accommodate multiple types of payloads and could be made sustainable so long as local governments and firms were involved in the operations.

Other companies are experimenting with different private-public partnership models to build local capacity for drone use and maintenance. The underlying idea is that these projects should not only entail Western countries flying drones over the developing world, using them as almost a laboratory for the technology, but rather they should produce a lasting benefit to the country itself through local partnerships and capacity building. Zipline, a Silicon Valley start-up, has developed a fixed-wing drone that can fly a radius of 80 km (49.7 miles) from its ground control station through all weather conditions (fig. 7.1). Equipped with GPS and an autopilot, Zipline’s drones can fly 150 km round trip powered by its lithium ion batteries without needing to be recharged.⁸⁵ They are relatively fast, flying at 100 km per hour, and can deliver parcels in as little as thirty minutes.⁹⁸⁶The drones drop packages using small paper parachutes rather than landing, thus obviating the need for runways at the drop point and limiting risk to the drones themselves. They can also take off and land with a catapult launch system that limits their physical footprint on the ground.

Figure 7.1 Zipline drone launch in a California testing facility, March 2018

The Zipline drone’s expected payload is small—1.8 kg, or roughly 3.96 pounds—but the intention is to deliver blood and emergency medical supplies to rural regions on a “just-in time” basis. Demand for certain medicines is intermittent in many rural clinics and keeping supplies of every medicine that might be needed in each rural health clinic is costly and inefficient.⁸⁷ Some medicines—for example, poison antidotes or snake-bite anti-venoms—may not be needed all of the time, but when needed, are urgently needed. Many medicines also have a short shelf life and can expire in storage before they are needed and used. It is obviously wasteful to simply hold them in stock at every location only to discover later that they cannot be used. To deal with these logistics problems, Zipline shortens the supply chain by allowing centralized health services to rapidly deliver medicine and blood on an as-needed basis. In 2019, Zipline has expanded its repertoire of delivered goods to include 170 different products, including anti-venoms, vaccines, and rabies treatments.⁸⁸

The first country that agreed to work with Zipline was Rwanda, which has a strong central government and a good health service, but like many countries around the world, has problems with delivering blood and other medicines in time to meet the needs of urgent cases. Most rural delivery in Rwanda is done by motorbike over dirt roads, which is expensive and cumbersome given the state of the roads in the mountainous country. Another advantage to working in Rwanda is its regulatory environment: the Rwandan government is aiming to make itself a technology hub of East Africa and has relatively few regulations on flying drones compared to more regulated areas like the United States and Europe.⁸⁹ Unlike many developed countries, Rwanda’s airspace is uncrowded and easy to use for drone test flights. Rwanda also has high levels of mobile phone usage in the country—one estimate suggests that 70%–80% of Rwandans have access to a cell phone—so this allows for rapid communication of needs from rural regions.⁹⁰

Zipline’s approach requires local doctors to send a text with details on the needed blood or medicine to a distribution center outside Kigali, which then prepares the package for the drone and deploys it on a preprogrammed flight to the specific location where it is needed.⁹¹ In 2016, Zipline began service by making 50–150 daily deliveries to twenty-one of Rwanda’s district hospitals, covering approximately half of the area of the country.⁹² By December 2018, Zipline was making deliveries to thirty district hospitals and 473 public health clinics around Rwanda and estimated that it was delivering more than 65% of all blood transfusions in the country outside the capital city of Kigali.⁹³ Its experience in Rwanda suggested that the local population was not hostile to the use of drones, in part because the Rwandan military had not used drones domestically and therefore the technology had few of the negative connotations it has elsewhere.

Zipline is clear that it is not a charity, but rather a for-profit health logistics company whose services are purchased by the local government.⁹⁴ This is very different from the traditional humanitarian approach in which drones and services are simply donated to countries in the developing world. This suggests that Zipline and other similar drone start-ups should not be cast in purely humanitarian terms, but rather as commercial endeavors designed to generate profits for their creators. There is nothing unusual or wrong with this; many other service providers in the humanitarian aid and post-conflict reconstruction worlds are also for-profit companies. But it does impose some additional burdens on the governments who make these bargains with commercial companies. One is to prove that the measure is cost-efficient: is this the best way for the Rwanda to deliver essential medical supplies? Or would it be more efficient to invest in building sustainable roads, rather than a high-tech solution? In part, the answer to this question hinges on how expensive it would be to build roads and other infrastructure in the developing world. It is not uncommon for countries in the developing world to have substantial deferred maintenance on existing roads and to struggle to afford that along with new infrastructure. For this reason, their governments may find that they get a bigger, immediate “bang for their buck” and cover more people with drone deliveries than they might with long-term, costly road improvements.

One country that has made this cost calculation in favor of drones is Ghana. In 2019, Zipline announced that it was partnering with the Ghanaian government to expand drone delivery operations on a larger scale than in Rwanda. Zipline aims to deliver 148 different medicines, vaccines, and other products to as many as 2,000 different health facilities across the country⁹⁵ and to build four distribution centers, each equipped with thirty drones, to enable this service. It estimates that it will make 500 deliveries a day from each distribution center, serving 12 million people.⁹⁶ Although the majority of its work is in Rwanda and Ghana, Zipline is not exclusively focused on Africa and plans to extend its operations to Latin America, North America, and other parts of the world. In each of these expansions, the rationale behind drone deliveries will hinge on complex calculations of cost and effectiveness against alternative options.

In the developing world, a particular challenge for drone deliveries is ensuring that these operations are sustainable over the long run. There is a danger with all of these projects that for-profit companies will simply use the airspace of the developing world as a laboratory, fine-tuning the technology for more lucrative markets like the United States while ultimately leaving the population there with nothing. Zipline has sought to address these concerns by working with Rwanda’s Ministry of Health and by hiring locals to do repairs, to operate the drone’s take-off and landing, and to process orders from local health clinics. If this investment continues in other locations and with other companies, it suggests that the deployment of delivery drones in a humanitarian context can be sustainable provided there is a commitment to develop the capacity of the local population to assume control over many, if not all, of the dimensions of a drone-based project.⁹⁷ If this capacity is built, it can also lead to entirely indigenous and self-directed drone delivery projects, as recently happened when Nepalese entrepreneurs and technicians developed a homegrown drone delivery service for medical supplies in the Himalayas.⁹⁸

Another project which straddles the humanitarian and commercial dimensions of delivery drones are the Redline cargo drones, hosted at the École Polytechnique Fédérale De Lausanne (EPFL) as part of their Afrotech program. As envisioned by former Economist journalist, novelist, and founder of Redline Jonathan Ledgard, fixed-wing drones would radically expand the delivery of cargo across the African continent, carrying packages of emergency medicine but also ordinary commerce weighing as much as 10 kg (22 pounds).⁹⁹Unlike Zipline, which aims to be cost-competitive with, and possibly replace, ground transport, Redline drones would be considered a supplementary system to the existing road and ground transportation networks. As such, they would not operate in a purely “last mile” context, but would rather fly from fixed, futuristic-looking drone ports that would be built throughout a particular country and would sustain the transport of medical and other commercial goods over a wider region. These drone ports, designed by the noted British architect Lord Norman Foster at the cost of $70,000 each, would become hubs of emergency supplies and later commercial goods, as well as refueling and repair stations for the drones themselves. The current Redline plan is to construct eighteen drone ports in Rwanda but eventually to move on to other, larger countries in Africa.¹⁰⁰At present, both types of drones remain in the prototype stage and the drone ports have not yet been built.

Redline drones would carry medical supplies, but its successor—the Blueline drone, capable of carrying a payload of 100 kg—would focus on commercial deliveries across underserved parts of Africa. Ledgard sees an enormous untapped potential in consumer spending in Africa among those who have cash but lack access to high-quality goods in major market towns because of problems with infrastructure and the postal delivery service.¹⁰¹ The ultimate goal of the Blueline drones would be to connect towns of between 20,000 and 50,000 people and to build a web of air cargo pathways that could facilitate the movement of goods and maximize the choices available to consumers. In his conceptualization, the humanitarian delivery drone (Redline) serves as a gateway for a wider commercial application of delivery drones (Blueline) that will stimulate economic growth across Africa. In the long run, Ledgard hopes that the drone ports themselves would become central hubs for drone and non-drone commerce in a particular region.

The Redline/Blueline model is an explicit effort to humanize drones and to recast them as something other than a purely military instrument. Drones have an inherently negative connotation because, as Ledgard notes, “there is a stickiness around the ‘drone’ word. If people were as scared of drones as they were a year ago, when there was a muddling in the public mind between military and civilian drones, there would be a problem. But people are now accepting that there are military and civilian drones.” At the same time, “the bristling word ‘drone’ which always conveys the idea of an insect, or of a worker doing a mindless task, is a problem. Intellectually, it’s better to think of this as flying robotics. But it’s hard in English to find a good word for this.” He has experimented with a number of other descriptions for cargo drones, including “flying donkeys,” in order to convince the local population that they are a normal, almost routine, part of commerce, rather than something to be feared. If the Redline/Blueline model succeeds, it will show the local population that medicine and consumer goods are just as likely to come from drones as are bombs. As Ledgard notes, the success of the Redline/Blueline model could be ultimately described as a “postmodern version of the idea of converting swords into ploughshares.”¹⁰²In cases like Rwanda, in which the post-genocide government has not only survived but emerged as a strong regional player, it is possible to see the investment in delivery drones in exactly these “swords and ploughshares” terms. But in cases where swords are still present—that is, in cases where countries have fallen into the abyss of civil war—delivery drones struggle to get off of the ground. To some extent, this is because delivery drones are hardest to make work in environments where they are most needed. As the Zipline example shows, some delivery drones projects presuppose a government that can work with aid givers, establish workable regulations, and exercise some control to de-conflict the local airspace. It is no accident that both Zipline and Redline have been tested in Rwanda, one of the strongest and most capable states in sub-Saharan Africa. These missions also presuppose trained staff, community engagement, and pre-mission planning, which often involves people on the ground in the area where the drops will happen.¹⁰³

In cases where political order has utterly collapsed due to civil war or other calamities, none of these preconditions are present, and as a result delivering humanitarian relief via drones becomes much harder. In 2014, Mark Jacobsen, a former US Air Force C-17 pilot, founded the Syria Airlift project to deliver medical supplies and water purifiers to besieged populations in the Syrian civil war, in which hundreds of thousands have died since 2011. His hope was that one team with ten fixed-wing aircraft could deliver 400 pounds of goods, in small packages, each night.¹⁰⁴ The project had some successful test runs, but ultimately failed at the pilot stage. The Syrian airlift project fell apart for a number of reasons, not the least of which was that requirements for safety and reliability for drones flying in such ungoverned, dangerous airspace increased the complexity and costs of the project. Few traditional backers, such as governments, were willing to assume the liability for funding delivery drones flying in airspace populated by US and Russian aircraft or controlled by the dictator Bashir al-Assad. Equally, for-profit companies could not be convinced relieving suffering through drone deliveries was somehow profitable.¹⁰⁵ The risks of delivery drones in these cases—that the packages might go awry, fall into the wrong hands, or be mistaken as bombs—were simply too high. With today’s technology, delivery drones are possible, but to work they need a favorable regulatory and political environment, a local population not immediately suspicious of the technology, and the alignment of the incentives of the project leaders and backers. It is unlikely that delivery drones will be a solution for reaching vulnerable populations in civil wars without significant improvements in the technology, high levels of confidence in their precision and reliability, and a significant drop in cost.

Peacekeeping Drones

While humanitarian organizations have been reluctant to deploy drones into conflict zones, the United Nations has been far more willing to countenance the use of drones in violent environments to keep the peace, although it is a relative latecomer to drone technology, only beginning to explore its potential in a systematic way in 2013. But the first use of a surveillance drone by UN forces was much earlier, in 2006, when European peacekeepers deployed to the Democratic Republic of the Congo (DRC) as part of the United Nations Organization Mission in the Democratic Republic of Congo (MONUC),¹⁰⁶ designed to monitor a cease-fire between various state-backed militias that had plunged the country into a bloody civil war in the late 1990s. The observation mission was composed of Belgian peacekeepers and a specially constituted European Force (EUFOR), both of which operated under a UN authority. The first drones came not from the United Nations itself, but from the national governments supplying the forces in the Congo. Both EUFOR and the Belgian forces used surveillance drones to interdict illegal arms shipments and had some success confronting the perpetrators with images of the shipments. The experiment ended after one drone was shot down and another was crashed in Kinshasa, killing one woman and injuring at least two others.¹⁰⁷ This crash also happened at a sensitive time, only a few days before a crucial election ratifying the post-war peace.¹⁰⁸

In 2009–2010, Irish troops deployed as part of a peacekeeping mission in Chad brought drones with them, which were used for surveillance. The situation was similar to that of the DRC: the drones came as part of the equipment belonging to a national contingent, as opposed to the United Nations itself, but were eventually put under the UN’s legal mandate for aerial surveillance.¹⁰⁹ The drones were used to monitor the movement of armed groups along the borders with Sudan and the Central African Republic and to improve the protection of refugees, internally displaced persons, and others.¹¹⁰ They were particularly useful in detecting incursions and threats to the civilian population when domestic opposition forces came across the border from Darfur in spring 2009.¹¹¹ The drones were widely considered successful, especially for making sure convoy routes were clear of bombs and other threats, but they had some limits when operating at night.¹¹² In Chad, the drones were capable of picking up images and relaying them back to the UN command but the absence of drones and helicopters capable of night flight slowed the response to threats like banditry and attacks on civilians.¹¹³

Despite these successes, the idea that the United Nations might use drones and other means to boost its surveillance capacity in peacekeeping made a number of governments particularly uneasy. Governments are typically jealous of their sovereignty and are reluctant to let anyone, especially from a foreign country, operate in their airspace without approval. In the DRC, Rwanda objected to the use of drones in part due to fears that UN drones would be hijacked by intelligence services of Western governments and used to expand surveillance in Africa and elsewhere. This is particularly important because information gleaned from drones could convey advantage in the tense geopolitics of the Central African region. Oliver Nduhungirehe, a Rwandan diplomat, argued that “Africa must not become a laboratory for intelligence devices from overseas . . . We don’t know whether these drones are going to be used to gather intelligence from Kigali, Kampala, Bujumbura or the entire region.”¹¹⁴ Some of the objections draw from the latent anti-colonialism that often distinguishes African politics and makes governments, like Rwandan government, suspicious of initiatives coming from Washington, London, and Paris. But some speculated that Rwanda, Uganda, and other countries were concerned that drones might reveal clearly the degree of their support for predatory militias in the DRC.¹¹⁵ Others were concerned that drones might eventually put peacekeepers out of work. For some countries in Africa and Asia, being a paid peacekeeper for the United Nations is a lucrative business, which they are loathe to lose to a new technology. But the reluctance to using drones in peacekeeping is not confined to Africa. Other countries, such as China, Guatemala, Pakistan, and Russia, also raised concerns over the use of drones in UN missions in Africa.¹¹⁶ Proposals to use UN surveillance drones in South Sudan fell afoul of similar concerns regarding information gathering and were outright rejected by its government.

Even with these concerns and objections, the momentum toward using drones in peacekeeping was difficult to stop, in part due to advocacy from within the United Nations to harness the technology for good. The first serious effort by the United Nations to support the use of drones was in February 2013, when the newly expanded United Nations Organization Stabilization Mission in the Democratic Republic of the Congo (MONUSCO) was granted the ability to buy and fly its own drone fleet. The first calls to equip MONUSCO with drones were made in 2008, when UN officials called for a surveillance capacity to boost their force protection and to allow for constant monitoring of the movements of government troops and rebel groups, as well as route clearance.¹¹⁷ But for years the debate was stalled over questions about the appropriateness of deploying drones in peacekeeping operations and the ownership of the images and intelligence coming from the drones. The obvious shadow hanging over this debate was the United States’ use of drones for targeted killing under the Obama administration. UN rapporteurs and human rights councils had traditionally been hostile to the Obama administration’s expansive policy on the use of drones, especially in countries like Pakistan and Yemen, where the United States was not officially at war. That lingering unease translated itself into delays and some restrictions on the nature of the United Nations’ drone use. The approval for drone use in the DRC happened in January 2013, but UN officials noted that it was contingent on the consent of the host government and that this authorization was “without prejudice to the ongoing consideration by relevant United Nations bodies of legal, financial and technical implications of the use of unmanned aerial systems.”¹¹⁸ The degree of unease around the United Nations’ approval of drones in the DRC shows how much the technology itself has been associated with its use in armed conflict and targeted killings. As a consequence of this, advocates of “good drones” in humanitarian relief and peacekeeping missions are often forced to engage in a sustained public relations campaign to sell drone use to the public, especially in war-torn environments and those where drones had been used previously for military advantage.

Under the leadership of Hervé Ladsous, the head of the United Nations Department of Peacekeeping Operations (DPKO), the UN Security Council approved the deployment of drones as part of a new and stronger mandate for MONUSCO’s in the DRC. Perhaps most important was the fact that the new mandate to use drones extended beyond protecting UN forces from attack and authorized the United Nations to use drones for the protection of civilians.¹¹⁹ Implicit in this mandate was the view that drones would be effective in deterring rebel groups from straying across borders or threatening civilians.¹²⁰ The United Nations initially planned to send three drones to the DRC for deployment in the northern Kivu province, but eventually pushed that number up to five. It selected the Falco fixed-wing drone, operated by the Italian company Finmeccanica, for the mission.¹²¹ Nearly half the size of a Predator, the Falco drone is capable of flying for ten to twelve hours at a time and at a distance of 150 miles. Operating at up to 18,000 feet, the drones are capable of seeing a wide swath of territory and capturing images of moving groups of people or trucks. Falco drones were equipped with synthetic aperture radar and a camera capable of taking nighttime and thermal images, as well as providing both still photos and real-time video.¹²² The capability to take good nighttime photos is crucial because it provides some degree of coverage for periods of time when no UN peacekeepers are normally available. A lack of infrastructure and poor roads in the DRC meant that UN peacekeepers generally patrolled in a limited radius of 10 to 15 miles from their home base and returned home at sunset. As a result, most militia activity, especially from particularly aggressive and deadly groups like the M-23 rebel movement, happened at night.¹²³ For the United Nations, providing some coverage at night was crucial because MONUSCO had come under fierce criticism for failing to stop or respond to night banditry.

The United Nations’ investment in drones in the DRC constituted approximately 1% of MONUSCO’s budget.¹²⁴ The original agreement paid a company called Selex $13 million per year to run and service its drones.¹²⁵ Each Falco drone flew between one and two missions per day of about five hours in duration. By some estimates, the drones were vastly more efficient than manned aircraft given that one UAV could cover about the same ground as fifteen to nineteen manned helicopters.¹²⁶ The use of drones in the DRC also freed up the helicopter fleet for other tasks and allowed UN peacekeepers to reposition themselves against greater threats.¹²⁷ There were also unsubstantiated claims that some rebel groups, such as the M23 and FDLR, who were backed by Rwanda, were more likely to surrender because of the pressure placed on them by drones. In DRC operations, drones were also able to spot capsized boats and illegal smuggling operations.¹²⁸But rebel groups hidden under a dense canopy of jungle were less likely to be affected by drones.¹²⁹

Figure 7.2 Technicians examine an unarmed UN drone in Goma, North Kivu province, Democratic Republic of Congo, 2013.

Despite these accomplishments, drones are not an unabashed success in the DRC. They are not in the air constantly nor are they capable of seeing a lot of rural areas where rebel group activity might be threatening the welfare of civilians. With often only one drone in the sky at a particular time, large parts of the DRC—itself a country the size of Western Europe—go unmonitored.¹³⁰ Even for those areas monitored, the imagery was returned too late to be considered useful, thus undermining the claim by UN Secretary General Ban Ki-Moon that drones would improve “timely decision-making.”¹³¹ Some critics suggested that the record of drones were overblown and that they had made “no discernible difference” to most ordinary people. At present, there is no systematic evidence to suggest that drone peacekeeping is effective in reducing violence. It is particularly difficult to assess the validity of claims that drones have been effective in deterring rebel groups from crossing borders or preying on the civilian population. Some UN officials are convinced that the mere appearance of drones is enough to worry rebel groups and to get them to rethink some of their actions, much like a thief would be deterred by a CCTV camera in a store.¹³² UN officials have admitted to flying Falco drones at low altitudes and exploiting the sounds they make to deter rebel groups on the ground.¹³³ They have also admitted to having drones hover over militant camps, as a way of saying “we know where you are, surrender.”¹³⁴ One UN official reports that that rebel groups have moved away from areas where drones are in the air.¹³⁵ If this is correct, drones may be altering the dynamics on the ground, but it remains to be seen if they are reducing violence. For example, the presence of drones in daytime might deter rebel groups from launching attacks when drones are watching, but this might only make nighttime attacks more likely. In other words, altering the dynamics on the ground does not necessarily imply a reduction of violence, only a redistribution of it. Actually measuring the effect of deterrence—that is, whether an overflight stopped an attack—is notoriously difficult to do and has not been attempted by the United Nations.¹³⁶ The idea that drones will deter attacks is intuitively attractive, but it is hard to know for certain whether these attacks have been truly stopped or merely re-directed toward times when the drones are not watching.

Even if the United Nations did get the images generated by the drones, it is not self-evident that they would respond quickly enough to make deterrence work.¹³⁷ The time that it takes to capture and relay images is a substantial problem, particularly for large geographic areas like the DRC. The images must be conveyed to a ground station and analyzed before they can be useful. Traditionally, the United Nations has been unable to do intelligence analysis in a systematic way because of internal reluctance to do something so explicitly “military”. It also struggles because national governments tend to be unwilling to share their intelligence with the United Nations or loan assets like drones to its forces. But the United Nations is moving in fits and starts toward developing a capacity for collecting and analyzing intelligence and in December 2014 published a report called Performance Peacekeeping that examined ways that the DPKO could use new and emerging technologies to enhance their ability to do peacekeeping missions.¹³⁸ Aside from recommending that drones be used for peacekeeping, policing, and border control, the report also recommended that the DPKO and other relevant bodies boost its capacity to collect data and images from a number of sources (satellites, sensors, CCTV cameras, and drones) to make peacekeeping a much more nuanced and information-rich enterprise. At the moment, this remains aspirational, as the UN mission in the DRC tended to have few analysts combing through the images, collecting and converting them into operational recommendations.¹³⁹ Only $13 million of the MONUSCO’s overall budget of $1 billion is currently spent on surveillance, and only a fraction of this is devoted to analysis.¹⁴⁰ By contrast, the UN mission in Mali has been able to use drones and to develop a more sophisticated joint intelligence center, led in part by European troop contributors with their own assets, than was possible in the DRC.¹⁴¹ But national governments who use drones are more likely to collect intelligence and share it with a secure circle of close military allies, as happened in Mali, than with the United Nations itself. Until the United Nations invests enough to develop the surveillance and analysis capability that national governments have, it is unlikely that their drones will be as comparably effective.

A more general problem with drone peacekeeping is that the presence of the drones themselves raises expectations in the civilian population beyond what the United Nations can currently achieve. To populations traumatized by war, the sudden appearance of drones can look like a technological miracle and suggest that the United Nations will be constantly watching and protecting them. The reality is very different: most drones currently only cover a limited area during a short period of time, and as such they can be evaded by adaptive rebel groups. Drones cannot fly effectively in bad weather or dense cloud cover and will crash or fail in sandstorms and other extreme weather. Even if drones spot incursions, it does not necessarily follow that the United Nations can or will respond with force to protect a civilian population. As the US record in Rwanda demonstrated, knowledge of an atrocity as severe as genocide does not imply that a government or organization will be moved to respond.¹⁴² Drones are a particular problem for setting expectations because as a sleek, futuristic technology they appear to promise too much. Some Congolese, for example, puzzled over why the drones were not armed and why they did not simply kill the rebel groups threatening them.¹⁴³ Setting expectations among the population about what drones can and cannot do—and by extension, what the organization possessing them is willing to countenance—is at least as important as the technology in the missions.

There is a further irony with expectations inside the United Nations itself. While raising the expectations of the population, drones can also increase pressure on UN commanders by raising expectations of what they can and should do in time-sensitive environments. A UN mission equipped with drones can see more and should make fewer mistakes than one in which information was received from other sources. For UN commanders, the flow of rich images from drones may add to the time pressure of their decision-making and raise expectations that they will make fewer, if any, mistakes. One danger pointed out by John Karlsrud and Frederik Rosén is that “radical improvements in knowledge accessibility will also lead to heightened requirements for the application of this knowledge when using force, as well as an increase in the demand for that force.”¹⁴⁴ UN commanders may find themselves being asked to do more to respond to images of atrocities coming from drones while at the same time to make fewer mistakes than ever before. The presence of video from drones means that they will be held accountable for mistakes in ways that would have been impossible before. In other words, drones may modernize peacekeeping, but also make it more demanding than ever.

Risks

The use of drones may be different in a number of ways for humanitarian operations and for peacekeeping, but it presents three similar risks. The first is privacy. From crisis mapping to watching borders in conflict zones, surveillance drones can capture images of people not just in public settings but also in homes, hospitals, and other locations where they have a reasonable expectation of privacy. Some of these drones will take only still images, but others will take live video and draw on other sources of data, including eventually SIM card data from cellphones, from targets on the ground. In emergencies like Nepal’s earthquake, it is obviously not possible to seek the consent of those filmed from above with drones. Equally, it is hard to control how the footage will be used. Especially following disasters in which new, often unregulated, actors appear on the scene with drones, there is relatively little information about who is filming individuals or what they will ultimately do with the footage. Similarly, when drones are handled by contractors, it is unclear what these private, for-profit organizations owe the local population in terms of privacy expectations.¹⁴⁵ As chapter 6 discusses, drones are scrambling the conventional expectations of privacy in the developed world, but they are also doing so in humanitarian crises in the developing world where the population is even more vulnerable. Some organizations are beginning to recognize this challenge and respond with guidelines. In January 2017, the Harvard Humanitarian Initiative published the Signal Code, which attempts to ground the protection of information in humanitarian crises as a human right and offers some guidelines for how responsible organizations should behave.¹⁴⁶ Other organizations like UAViators have also begun to develop codes of practice which lay out the expectations of privacy that individuals on the ground should have in humanitarian operations. The problem remains one of enforcement: if an organization violates the privacy of an individual during a humanitarian or peacekeeping mission, who can punish them effectively? How can we protect the privacy of people in desperate situations whose names we may not even know?

Another set of risks concerns who owns the data and images that are produced as a result of drone overflights. The guidelines and procedures for retaining and sharing information and images in crisis zones remain unclear and somewhat contradictory. Many of the “digital humanitarians” flying private drones over places like Nepal and Haiti have argued that data and images should be shared widely and even uploaded to common software like OpenStreetMap in order to make relief efforts easier. Although strictly speaking the images are owned by the organization that flies the drones, the cultural emphasis of the digital humanitarian movement is toward open source sharing so that as many people as possible can see and help those in need. But an absence of strong regulations on sharing images and data from drones presents a problem because freely shared images could be misused by some governments to hunt enemies, attack dissidents, and put pressure on refugees to leave. In other words, the digital humanitarian movement needs to be careful about simply throwing images up online with the assumption that only people who want to help will see them. As responsible organizations like UAViators acknowledge, this movement needs instead to develop data and image protection policies that acknowledge how these resources could be misused for malicious purposes. The United Nations is slightly different: it owns the data and images drawn from the drones and is keenly aware of the potential risks of misuse. As a result, the United Nations does not always share what it collects even with the host government or partners in its operations. This can cause friction with some host governments who have claimed that the United Nations is collecting intelligence on security threats which it refuses to share. But the United Nations has its own problems in terms of data ownership. As an institution, the United Nations has traditionally lacked a clear policy on privacy and has weakly enforced the handling of classified data and images inside some missions.¹⁴⁷ The absence of clear regulations and guidelines for the use of data and images from UN drones will only multiply the policy dilemmas that commanders face on the ground. For example, it remains unclear whether the United Nations can or should share data with local police forces or whether evidence that it collects could be turned over to the International Criminal Court.¹⁴⁸ Transparency about what is collected, and who may see it, remains a serious concern for UN member states.¹⁴⁹

Finally, the humanitarian use of drones has itself been complicated by UN efforts to use drones for peacekeeping purposes. Those deploying drones in humanitarian emergencies for mapping or search and rescue have always framed their drone deployments as consistent with the principle of neutrality. Such deployments are committed to helping all and delivering information and aid to everyone in need rather than only one side in the conflict.¹⁵⁰ This principle of neutrality is well established in humanitarian practice and enshrined in UN resolutions and the practice of the International Committee of the Red Cross (ICRC). For many digital humanitarians, it is an article of faith and a principle by which the legitimacy of their operations is conferred. By contrast, a number of recent UN peace enforcement missions are explicitly non-neutral and involve the United Nations taking a side and punishing the parties seen as most threatening to the peace. In the DRC, for example, UN forces fought with government forces against the M-23, FDLR, and other rebels seen as dangerous to the peace; in Mali, UN actions were directed against Islamist militias threatening the government. In these cases of peace enforcement, UN surveillance drones may not be armed but they are certainly working for one side in the conflict. For this reason, a number of humanitarian organizations are concerned that the use of drones by the United Nations for peacekeeping might mean that “the lines are blurred between help and harm,” or between neutral and non-neutral actors in a conflict.¹⁵¹ One consequence of this might be that rebel groups will fail to distinguish between neutral humanitarian drones and non-neutral UN drones and attack the former. Others worry that humanitarian organizations on the ground will be targeted in reprisal for military activities that the drones of the United Nations or national governments are involved in. In a world where a number of actors are deploying drones in conflict zones for different reasons, drawing a recognizable distinction between humanitarian drones, UN drones, and even local military drones is increasingly important. One simple potential remedy for this is to paint drones in different colors to ensure that they are distinguished as humanitarian, as opposed to military or UN vehicles. But even this will require the local population understanding the difference and tailoring their responses to the drones accordingly.

Conclusion

The use of drones for “dull, dirty, and dangerous” missions has been both lauded and criticized for providing a positive face for a technology that is more closely associated with military use and targeted killings.¹⁵² There is certainly evidence that drones can do good in humanitarian and peacekeeping missions and bring capabilities for surveillance and data collection honed in military operations to bear in innovative ways. At their best, they may be able to “lift the fog of war” that makes delivering humanitarian relief and establishing a workable peace so difficult.¹⁵³ But it is important to measure the risks of their use in these environments, rather than just succumb to techno-utopianism about their potential effects. Drones will not replace aid workers or revolutionize aid delivery in the short term. Even the strongest advocates of humanitarian drones acknowledge that the technology must improve and the costs of drone use must plummet in order for the impact of drones will be known.¹⁵⁴ It is also important that the organizations’ goals for drones remain limited: it is possible, in an environment of overwhelming need, for goals to expand and for drones to be attached to all manner of problems, even when they are not the best tool to use.

Perhaps a greater risk of drones in this space is that of virtual humanitarianism. The movement toward disaster drones has produced a rush of new actors into crisis areas, but many are start-ups whose interest in the country extends only for a short period of time. Their surveillance and airdrop activities, however useful in the short term, are not always certain to produce a long-term benefit for the population, especially as global attention shifts to the next crisis. The danger is that humanitarian crises and peacekeeping missions could become little more than laboratories to test drone technology on a population in stark need but with little agency about how it is treated. Drone humanitarians must guard against the belief that they can simply provide data or airlift parcels of food and medicine by drones and be done with a crisis. Such an approach might be a natural temptation with unmanned aircraft, but if left unchecked it will undercut the genuine humanitarian purpose of the mission. For both humanitarian crises and peacekeeping, it is important that drones not be directed simply toward managing the effects of a crisis, but rather be embedded in a comprehensive approach that is directed at ending political instability and alleviating human suffering over the long term.