To fly drones over roads safely, study what would happen if they crash
A drone slams into a windshield. The drone's shell cracks, and the windshield shudders. Pieces of a propeller snap off. But the windshield doesn't shatter.
Most safe drone flights aren’t supposed to end in crashes. But this crash is unfolding in one of Virginia Tech’s impact labs, where controlled collisions have been pivotal to producing safer cars, football helmets, and even toys. Here, collisions reveal information about risk and point to ways to reduce it. Now they're helping lead to new ways to use drones.
For years, an important Federal Aviation Administration (FAA) regulation restricted commercial drone use in urban and suburban areas by prohibiting flying over moving vehicles. The concern was that a drone hitting the windshield of a moving car could obliterate the driver’s view of the road or, worse, injure the occupants.
Data demonstrating that a small parachute-equipped drone wouldn’t actually damage a car traveling at typical in-town speed limits provided a counterpoint to those concerns – one convincing enough to the FAA that it granted insurance company State Farm a waiver to operate drones over moving vehicles in 2019.
The data represent the first known example of direct research on drone-automobile collisions – a unique collaboration between drone specialists at the Virginia Tech Mid-Atlantic Aviation Partnership (MAAP), injury biomechanics experts in the university’s College of Engineering, and State Farm.
The waiver provided the company with an additional tool to use in processing homeowners insurance claims.
At the end of 2020, the FAA issued new rules that expanded permissions for drones to fly over moving vehicles. But the more restrictive rules in place when State Farm and MAAP started working on the project would require a drone operator assigned to inspect two houses across a busy street from each other to finish the first inspection, land, walk the drone across the street, and start all over just a few feet away. The waiver allowed the drone to fly from one house to the other instead.
State Farm has been working with MAAP for several years to safely incorporate drones into its business operations. As an FAA-designated drone test site, MAAP is one of a handful of groups leading research that will guide the safe integration of drones in the United States.
“State Farm claims inspections are the kind of operation where drones can offer huge advantages,” said Mark Blanks, MAAP’s director. “They also required a meaningful expansion of the existing regulations. The intersection of those factors — an opportunity to help realize the potential of this technology that will also advance the industry as a whole — is exactly where our mandate lies as a test site.”
In the aftermath of the 2018 hurricane season, the partnership between State Farm and MAAP yielded the first nationwide waiver for drone operations over people. This new waiver was another industry first.
“State Farm is committed to finding innovative opportunities that support safety and efficiency in our claims handling,” said State Farm Manager Todd Binion. “If we look for ways to use data to help overcome regulatory challenges, we can move forward in other ways where drones can be used to benefit consumers across multiple industries.”
Getting a waiver that’s never been granted before, for technology that’s rapidly evolving and being used in new ways, means finding ways to define, quantify, and categorize risks in new and novel ways.
“What the FAA wants to see is a compelling, credible case that the operation is safe,” Blanks explained. “You need extensive data that’s grounded in the context of how the operation will work in the real world. There wasn’t any empirical information on what would happen during an impact between a drone and a car, so in order to craft a strong waiver application, we had to come up with research methods to obtain that information.”
MAAP regularly collaborates with Virginia Tech’s acclaimed injury biomechanics researchers. The group, led by associate biomedical engineering and mechanics professor Steve Rowson, has earned international recognition for experimental methods that precisely recreate real-world collisions. These techniques offer a realistic and, crucially, reproducible way to quantify the risk of various impacts and evaluate the efficacy of safety measures.
The drone the team used was DJI Mavic 2 Pro, a small aircraft that weighs just over 2 pounds. State Farm’s waiver application specified that during operations the drone would be equipped with a parachute from Israeli company ParaZero, technology that has been tested against ASTM standards and has figured in multiple successful applications for waivers to fly over people. The researchers set out to replicate what might happen if, in the unlikely event of a catastrophic failure, the drone under the parachute was blown directly into the path of a moving vehicle.
They mounted the drone to a pneumatic sled that was propelled down a linear track toward a custom-built windshield frame that contained parts of a salvaged Toyota Camry. By varying the impact speed - which would translate to the speed of a moving vehicle in a real-world scenario - and assessing the damage, the team hoped to identify a threshold below which a drone could safely be flown overhead without risking a smashed windshield if something went wrong. (A similar setup, with a crash test dummy at the business end of the track, had supported their earlier research on drone impacts with humans.) They expected the degree of damage to the windshield to increase steadily as the impact speed rose.
Post-impact inspections showed a more interesting result. At impact speeds between 25 and 62.5 mph, the only signs of the collision were streaks of rubber and plastic transferred from the drone as it slid up the glass. When the Mavic hit at 67 miles per hour, a web of cracks shot across the windshield as it bowed inward, spitting shards of glass onto the floor.
The slim margin between a virtually pristine windshield and a destroyed one drew a clear boundary around low-risk scenarios. The waiver application argued that as long as potential relative impact speeds never exceeded 62 mph, flights over moving vehicles presented minimal risk. The FAA agreed.
Using the waiver in the field takes some consideration of the environment. The speed at which a drone would hit a windshield is the sum of the vehicle’s speed and the speed of the wind pushing the drone toward it. In the average suburban neighborhood, with speed limits between 25 and 35 mph, even the maximum wind speed the drone can tolerate — about 23 mph — would still keep potential impact speeds below the 62 mph threshold.
The FAA’s more permissive rules for flights over moving vehicles are slated to take effect this spring. However, Blanks said, research like this will continue to be valuable for companies pursuing the highest standards of safety.
“If you're dedicated to thoroughly understanding and mitigating risk in their operations, having direct, quantitative data about the consequences of an impact is going to remain relevant,” he said. “The fundamental principles and test methods have provided insight for standards development and rulemaking, and they’ll continue to be part of the foundation of safety that’s going to be crucial to this industry as it matures.”