As countries around the world establish drone regulations, they seem to have settled on using a mass of 250 grams, or about half a pound, as the threshold to require registration and/or further permission to fly. That covers all but the smallest of consumer toy drones.
But a research report from DJI, the world’s largest drone company, suggests that weight is far too conservative — by a factor of almost 10x.
The report, embedded below, is technical and thorough (and, to be sure, self-serving). But it offers an interesting glimpse into how decades-old assumptions are sometimes used — perhaps incorrectly — in modern policy making.
The 250g threshold was decided during a U.S. Federal Aviation Administration summit in November 2015; participants included Amazon, Alphabet’s X “moonshot” subsidiary, Walmart and DJI, among others. The goal there was to come up with a U.S. registration criteria for low-risk drones in a short amount of time.
In this research report, first published in December 2016, DJI analyzes the variables that went into calculating the 250g limit. These include the shape of the drone, its likelihood of failing, the population density of an area and the likelihood of death if a falling drone hits a person. For each variable, DJI walks through the range of possibilities considered by the summit — from optimistic to pessimistic — aiming for the most realistic result.
One of the most interesting things it uncovers: One of the variables, the probability of fatality from getting hit by an object, has a rather long history, and ultimately dates back to an estimate derived from a 1968 computer simulation on the dangers of nuclear war. (To figure this out, the report traces back through almost 50 years of citations, starting with a chart from a 2012 study, which then cites a 2010 report, referring to data from a 2007 paper, citing a study from 2000, that ultimately goes back to the 1960s.)
In that original simulation, researchers were trying to evaluate the effectiveness of shelters in the event of a nuclear attack, figuring out the probability of injury or fatality through various fallout scenarios.
As DJI describes:
The nuclear casualty report itself relied on previous studies involving shooting balls and other impactors at animals, force to crush cadaver skulls, and other actual data relating various kinds of trauma to injury. The authors realized these data had a basic problem, however. Most of the studies for impact, for example, had been made using fairly light, compact objects, like small steel balls or cubes. Their study needed to extrapolate from the data they had to the effects of much larger impacts, such as might happen when a building is blown into a person’s body during a nuclear blast. So the authors created curves with data they had, which involved data from impactors less than 10 grams in mass, extrapolating to estimate the effect from impactors ranging from ~1 gram to ~50 kilograms. Their simulation needed to account for very small impacts from shattered windows, up to big impacts from chunks of concrete buildings.
Those led to the lethality curve — showing the probability of death after getting hit by an object carrying a specific amount of kinetic energy — that was eventually used in the FAA’s 2015 drone safety equation.
The problem is that the context of full-scale nuclear fallout simulations might not best represent modern everyday life. The calculations “were based on assumptions dating back 48 years about the lack of medical care in a thermonuclear war,” DJI notes. For example, the researchers figured that getting cut with glass carried with it a 10 percent chance of death — which now seems highly improbable.
This makes “any reliance on the curve for expansive [drone] policymaking appear ill-conceived,” DJI writes.
DJI proposes a more realistic case for estimating the likelihood that a falling object will kill someone: A 2004 Department of Justice report on the the effects of being hit by non-lethal munitions, like bean bags, which are used for crowd control by law enforcement.
Other assumptions, such as those about the shape and aerodynamics of a drone, are potentially more problematic. There, the task force assumed that a drone is shaped like a race car, which has very little drag and will pick up a lot of speed before it crashes, Brendan Schulman, head of policy at DJI, told Recode in an interview. But a typical quadcopter drone is shaped more like a cube, which is more drag-inducing, Schulman said.
More broadly, the fact that any of this can be traced back to the 1960s suggests that more research could be done — and that’s ultimately what DJI wants — before this specific weight is accepted as the global standard for drone regulations that range from simple registration to more stringent restrictions.
In Canada, for example, a proposal is circulating that would require pilots flying drones over 250g to take a knowledge test and obtain insurance. In Russia, the 250g weight limit is also used as a threshold for registration. And in the European Union, 250g is being proposed to delineate a category of low-risk aircraft, too, where any drone heavier than that has to be registered and can only be flown by a pilot who is at least 14 years old.
Changing these sorts of variables leads to a much different result: Through its analysis, DJI concludes that a mass limit of around 2.2 kg would be the “appropriate” threshold for drone registration. This, then, would put most modern, consumer or “prosumer” drones outside the threshold. For example, DJI’s newish Mavic drone, which weighs in around 740g, currently requires registration — but might not if mass limits are loosened. (Update: A DJI rep clarifies that the company is not challenging the 250g cutoff as a registration limit, but rather as a safety standard in policymaking.)
Is that good? It would obviously be helpful for DJI and other companies that hope to sell a lot of drones with minimal market friction. On the other hand, it also seems like a good idea to require drones that have the ability to fly beyond an operator’s line of sight to be registered; if anything, it could help hold accountable people who fly drones in a reckless or dangerous way. And it’s not like drone registration has severely limited the market: The FAA reported that more than 500,000 drones had been registered in the first eight months of its program.
Still, even those in attendance at the 2015 FAA meeting expressed reservations about using the 250g figure. The registration system launched just a little over a month after the meeting, and some participants “felt there was insufficient time afforded to fully evaluate the calculations and assumptions made that resulted in the 250g cutoff weight,” according to a report the group issued afterward.
For what it’s worth, the FAA seems happy today with the 250g limit, regardless of how it was reached.
“The exact safe weight of an unmanned aircraft falling on a person is subject to many factors, including its physical configuration,” an FAA spokesperson told Recode. “Based on testing to date, we are confident the 250g weight selected represents a safe weight.”
Here’s the full report from DJI:
Defining a Lowest-Risk UAS Category (1) by April Glaser on Scribd
This article originally appeared on Recode.net.