Sections

Commentary

How to solve the problem of missing drone crash data

A drone helps give the Indianapolis Fire Department an aerial look as they respond to an industrial accident on Thursday, June 3, 2021.

For drone pilots, danger is everywhere. An undertrained or unlucky pilot can screw up, press the wrong button, and cause her drone to plummet to the ground. A light drizzle might permeate a pilot’s drone and send it rocketing into a tree. Almost everyone who flies drones for a living has witnessed at least one such an incident, in which a drone crashes, is lost, or otherwise malfunctions.

Despite the certainty that drones will crash—and perhaps endanger public safety—it’s all but impossible to determine how often such incidents occur in the United States, where there’s little publicly available data on drone crashes. This lack of data applies not only to civilian crashes, but also to the drones that are flown ever more regularly by government entities, like police and fire departments. While drone crashes don’t seem to be terribly common—and there are still no known cases of a small drone crash killing anyone—it’s still important that pilots, regulators, and the public have some sense of how often they happen and in what patterns they occur. With better data, regulators could more quicky identify unsafe practices and badly run drone programs, control malfunctioning, and identify badly made equipment.

The UK’s approach to drone crash data

Some countries are doing a much better job of tracking and reporting drone incidents than others. In the United Kingdom, for example, members of the public have quite a bit of data work with to understand the likelihood that a drone might come down on their heads. There, the UK Air Accidents Investigation Branch (AAIB) releases a steady tempo of public investigative reports describing incidents involving small drones. Per the AAIB, regulation EU 996 dictates “any person involved” with knowledge of an aircraft accident or serious incident in the UK must report it to the organization. The criteria for what determines an Unmanned Aircraft System accident are very broad and can be interpreted as applying to any accident where damage to the aircraft extends beyond damage to the propeller blades alone. Pilots are urged to report on accidents even if they are in doubt about whether it ought to be reported. The AAIB publishes crash reports in monthly bulletins, the most recent of which, from September, includes 19 incidents in which drone pilots got in touch to report a mishap.

This data makes it easier to detect problematic patterns. Consider the case of the expensive DJI Matrice 210, a rugged drone that’s popular among government agencies around the world. Between December 2017 and October 2019, the UK’s AAIB logged 16 different accidents or incidents involving a Matrice 210. In 13 of these cases, police or emergency services were behind the controls. It’s a less-than-inspiring safety record, prompting the AAIB to warn pilots about the possibility of moisture causing them to lose control of their aircraft.

While there’s significant debate whether these failures were due to technical issues or unrealistically high expectations of the drone’s performance in rain, the UK media pounced on the story, poking fun at public safety agencies who—in a famously rainy country—purchased drones apparently prone to malfunctioning when exposed to damp conditions. The availability of AAIB data makes reporting on drone mishaps a regular feature of UK media, as in January 2021 when a police drone pilot with just 6 hours of training was, inexplicably, given full control of a an almost $70,000 Aeryon SkyRanger drone. He pressed the wrong button and sent it plummeting to the bottom of a pond.

A paucity of data on U.S. drone crashes

In the United States, on the other hand, there are far fewer sources of information both in media and government sources about drone safety incidents. While stories about public safety drone mishaps regularly pop up in UK media, they’re almost nonexistent in the USA. My extensive recent search for such articles yielded only handful, mostly from the 2010s, years before it was broadly legal for public safety agencies to operate drones. Government figures on U.S. drone crashes are almost non-existent or fragmented.

This lack of data on drone crashes isn’t because drones aren’t being flown by civilians and police departments in the United States. According to 2020 figures compiled by the Center for Research on the Drone, at least 1,578 public safety agencies (most of which are police) now own their own UAS. That means that police drones have been acquired in large numbers. Presumably, most are being flown, at least on occasion. And yet there seems to be a curious cone of silence, both in the media and in official records, around any crashes or malfunctions that might take place during those outings.

Let’s again consider the case of the not-so-rain tolerant DJI Matrice 210. DJI products—and the Matrice 210 in particular—are very popular in the United States. While DJI doesn’t publish sales data publicly, analysts estimate that the company alone controls 90% of the North American drone market. They’re also very popular with public safety and government drone operators: The Center for Research on the Drone found in its 2020 study that a large majority use DJI drones, as opposed to those made by other drone-makers. We can assume that DJI Matrice drones sometimes malfunction or break for reasons that aren’t the fault of the operators. We also can assume that, at least occasionally, it rains in the United States. We can also assume that public safety drone pilots in the United States, just like those in the UK, occasionally do foolish things. Surely, then, there must be some reports of American Matrice 210s coming to an unpleasant end under similar circumstances as those in the UK. And yet, my searches found only two publicly available reports—a local newspaper story and a blog post—of U.S. crash incidents involving public safety pilots and DJI Matrice 210s.

If one was to simply examine the public record, you might conclude that American public safety drone pilots are brilliantly skilled with superb luck and never experience technical malfunctions or inclement weather, unlike their counterparts in the UK. Of course, that’s nonsense. Simple logic and the law of numbers dictates that public safety drones will occasionally suffer mishap and misfortune. As the United States is a big country with lots of public safety drones, it follows that there would be even more such incidents than in the UK. And yet, in the United States, we just don’t seem to hearabout them much.

There are a few reasons for this U.S. discrepancy. Currently, drone makers and the government aren’t required to disclose safety issues to pilots, nor are they required to safety test their drones before selling them to the public. At the moment, there’s no mandated safety-testing or airworthiness protocol imposed upon civilian drone makers in the United States, though last year the FAA published airworthiness criteria that may make it possible to certify drones produced by 10 different commercial drone makers.  

While some drone crash incidents must be reported to the U.S. government, the current rules only apply to a fairly small swath of incidents, as compared to the much broader interpretation under UK law. Under U.S. law, only certain types of non-recreational drone incidents—including those involving drones weighing more than 300 pounds, property damage costing more than $500, certain kinds of flight control system malfunctions or failures, and incidents that cause serious injury or death—must to be reported to the FAA and the NTSB. Pilots who are involved in an incident can also voluntarily file a report to the Aviation Safety Reporting System (ASRS), which is run by NASA.

Unfortunately, these databases contain very little publicly available information on crashes involving small drones. A search of the NTSB’s CAROL query tool for post-2008 incidents resulted in only 5 hits involving drones operated under the FAA’s Part 107 rules (which regulate all commercial, or non-hobby, drone uses, including public safety and government operations). There were only 21 hits that involved any kind of unmanned aircraft at all (mostly large and/or experimental). The FAA website publishes preliminary incident data over the last 10 days—then refers you back to the NTSB. A search of the ASRS database returned a mere 14 reports from between 2019 and 2021, none of which involved public safety.

This paucity of official sources of information helps explain why there are so few news stories about drone incidents. Public safety agencies that crash their drones often aren’t obligated to report the incident to the public (and often aren’t required to report to the FAA, either, under the existing criteria). In the absence of these reliable sources, journalists don’t have much to work with.

The best currently available resource for U.S. drone incidents appears to be a privately run website called Report Drone Accident that collects user-submitted reports on drone crashes and incidents. Pilot Steve Rhode was inspired to launch the site after he suffered a scary, potentially dangerous drone malfunction while flying a Matrice 210 during a search and rescue and mission. After that incident, he searched for reports on similar Matrice 210 incidents, could find no centralized information on the matter, and set up the site. Of its more than 300 entries so far, at least 40 concern the DJI Matrice 210. While Rhode thinks this is as much a function of user error and the drone’s popularity as it is fault on the part of DJI, it’s still a worrisome statistic. “Drone manufacturers aren’t competing on safety—they’re not aircraft manufacturers, they’re just selling a widget,” Rhode says. “Responsibility is all up to the end user, the pilot.”

Improving data access

What can we do about this lack of information on how often civilian drones, including those used by police, fall out of the sky? The legal situation is changing rapidly. In late July, the NTSB issued a notice of proposed rulemaking that would change the existing criteria for drone crash reporting. Currently, drone pilots are only required to report crashes to the NTSB under fairly specific, weight-based criteria. For drones that weigh under 300 pounds, pilots are only legally mandated to report incidents that result in “death or serious injury,” while pilots must report all incidents involving “substantial damage” to the drone itself if the aircraft weighs over 300 pounds.  

From the NTSB’s point of view, this system is outdated in the era of ever-more frequent lightweight drone flights in populated areas. The proposed rules would eliminate the weight-based definition. Instead, it would use airworthiness certificates as the criteria: If a drone of any size holds an airworthiness certificate and suffers “substantial damage,” then the NTSB will be empowered to investigate the situation. While the NTSB’s authority here wouldn’t apply to drones without airworthiness certificates, the agency assumes (almost certainly correctly) that certificate-less drones will become an increasingly rare liability in the near future, as regulators will likely require certificates to exempt drones from certain restrictive rules, like flights at night or beyond visual line of sight.

Hopefully, these new NTSB definitions of what constitutes a report-worthy drone accident will translate into a lot more reports on small drone crashes in the existing public NTSB database. That would make the currently drone-data-sparse database a much more valuable resource for everyone seeking real-world information on drone safety. Two drone industry organizations, the Small UAV Coalition and the Association for Unmanned Vehicle Systems International (AUVSI) broadly favor the changes but propose adding language to the NTSB’s definition that mirrors the FAA’s Part 107 rules, which only trigger an accident investigation if the cost of drone repairs or property damage exceeds $500. However the final language shakes out, the NTSB’s proposal would be a step in the right direction toward ensuring that the drones used in public safety and for commercial purposes are safe and that issues are well-documented while still leaving some wiggle room for amateur and non-commercial drone pilots to experiment with new aircraft.   

Another common-sense reform would place stricter requirements upon drone users in positions of responsibility who fly in higher-risk situations, like pilots who fly drones in populated areas on behalf of public-safety agencies. A police officer who intends to regularly operate drones over crowded public areas and protests should be held to stricter crash reporting standards than a private sector or Forest Service drone pilot who only flies over isolated areas where very few people are present. By collecting more detailed data on public safety drone uses (especially higher-risk operations, like flights over people), we’ll have a better sense of which drone flight operations are “worth it”—and we’ll know when it’s better for drones to stay on the ground. Ideally, this data would be fed into a publicly accessible database like that operated by the NTSB, and would be easily searchable by reporters, researchers, and drone industry professionals.

Drones offer immense value to private industry and the public sector, but we also need to be confident that the drones that are flying in ever-increasing numbers in our skies are as safe as they can possibly be. As the drone industry matures, it’s high time that the public knew more about the likelihood of a drone falling out of the sky and into their pool—or on their head.

Faine Greenwood is a consultant and writer on civilian drone technology.

The Brookings Institution is committed to quality, independence, and impact.
We are supported by a diverse array of funders. In line with our values and policies, each Brookings publication represents the sole views of its author(s).