Quadrotors have a reputation for being both fun and expensive, but it’s not usually obvious how dangerous they can be. While it’s pretty clear from the get-go that it’s in everyone’s best interest to avoid the spinny bits whenever possible, quadrotor safety primarily involves doing little more than trying your level best not to run into people. Not running into people with your drone is generally good advice, but the problems tend to happen when for whatever reason the drone escapes from your control. Maybe it’s your fault, maybe it’s the drone’s fault, but either way, those spinny bits can cause serious damage.
Safety-conscious quadrotor pilots have few options for making their drones safer, and none of them are all that great, due either to mediocre effectiveness or significant cost and performance tradeoffs. Now researchers at the University of Queensland in Brisbane, Australia, have come up with a clever idea for a quadrotor-safety system that manages to be highly effective, reliable, lightweight, and cheap all at the same time. If that sounds too good to be true, we have video of some hot dogs not getting chopped into bits that might convince you otherwise.
The safest quadrotor that we can think of is probably Flyability’s Gimball, or one of the other drones that uses a wraparound cage. It’s very effective, but comes with a significant size and weight penalty, and it’s particularly annoying for drones with cameras. It’s possible to scale the cages down to completely enclose just the rotors, but that’s potentially just as expensive and also much less aerodynamic. All of these safety systems are passive; active safety is also an option, but then you have to worry about things like sensors and computers always working reliably, along with costs that can escalate quickly.
The ideal drone safety system would provide reliable protection from any obstacle approaching from any direction, and it would do so with a bare minimum of cost and weight. Reliable means that the system really needs to be passive, but it also needs to offer the same protective coverage as a rotor cage without all of that added mass. The University of Queensland researchers have developed a system which meets these criteria, in the form of a swept mechanical interference sensor called the Safety Rotor. It’s a simple idea: A plastic hoop is added to the rotor system that spins around the rotor plane, such that anything that would make contact with the rotor must make contact with the hoop first. And if the hoop senses a contact, it puts the brakes on the rotor, slowing it enough that it’ll turn needing a finger into needing a band-aid.
While the concept here seems simple, the details are what makes it practical. The hoop spins passively, driven by a small amount of friction against the rotor hub that causes it to rotate at a few tens of hertz. This is fast enough for quick obstacle detection, but slow enough that the hoop itself isn’t a danger. The base of the hoop is studded with IR reflectors, which pass in front of an IR detector mounted near the rotor hub. All the system has to do is make sure it keeps getting consistent pings from the detector, and if too much time passes between pings, it means that the hoop has run into something, and the system engages the rotor brake.
The brake itself is electrodynamic, and functions by essentially shorting the motor inputs to turn it into a generator instead. The current generated by the spinning motor opposes the direction of rotation, and the faster the motor is spinning, the stronger this negative torque is. It would be marginally faster to apply voltage to actively decelerate the motor, but that would make the system more complicated and introduce more things that could go wrong. Another advantage of keeping things passive is that total loss of power will cause the braking circuit to activate, as will a safety system that isn’t set up properly.
Image: University of Queensland
To get a sense of how well the Safety Rotor works, let’s first look at the cost— what you have to pay to get it working in the first place. This can be measured in a bunch of ways, but the first is with money, and the good news is that the researchers estimate that integrating four Safety Rotors on an average drone would cost just over $14. It’ll also cost efficiency, but not all that much: Total weight added is under 22 grams, and aerodynamic effects don’t seem to be significant. The end-user also has to pay with their time and effort to set the system up and keep it running, but it’s quite simple, and doesn’t require skill to operate once it’s installed.
For that cost, what do you get? The researchers did some experiments, and this is what they found:
The measured latency [of the Safety Rotor's braking response] was 0.0118 seconds from the triggering event to start of rotor deceleration. The rotor required a further 0.0474 s to come to a complete stop. Ninety percent of the rotational kinetic energy of the rotor (as computed from angular velocity) was dissipated within 0.0216 s of triggering, and 99 percent of the rotational kinetic energy of the rotor was dissipated within 0.032 s.
The safety functionality of the safety system was tested on the bench using a processed meat “finger” proxy to trigger the hoop, and also applied to an open rotor (without hoop) for comparison. The rotor was spun at hover speed (1100 rads−1) and the finger proxy was introduced into the hoop at 0.36 ms−1 … The rotor and finger motion were captured using a shutter speed of 480 Hz. The rotor came to a stop within 0.077 s, with only light marks on the finger proxy from the impact of the hoop. The rotor was completely stopped by the time the finger reached the rotor plane. In contrast, the tip of the finger proxy introduced to an open rotor was completely destroyed.
The faster the quadrotor is moving, of course, the less effective the safety system will be, since the time between hoop contact and rotor contact decreases and there won’t be as much of a chance for the rotor braking to take effect. Performance will be highest during low-speed maneuvering, take-offs, and landings, but these are situations in which quadrotors seem to present the most consistent danger to both operators and bystanders. And again, it’s really a question of how much benefit you get in exchange for the cost.
Fifteen to $20 for a safety system with this level of performance is super cheap relative to the price of most consumer quadrotors, especially when you consider that add-ons like simple propeller bumpers cost $10 to $20 all by themselves, and more comprehensive safety systems like propeller cages can run well over $100. Even if manufacturers were to make this kind of safety system an option that consumers could electively bear the entire cost of, it seems like it would be an easy choice for anyone just starting out, considering how significant the upsides are and how insignificant the downsides seem to be. Personally, I’d happily pay it—if it never comes in handy, I’d be out $20, but if it saves even one finger, that’s priceless. Or it’s however much a new finger costs nowadays, I guess.
For more details, we spoke with Paul Pounds via email.
IEEE Spectrum: Why do you think someone hasn’t come up with a safety system like this before?
Paul Pounds: It has been difficult enough making quadrotors fly for useful lengths of time, that heavy and expensive systems haven’t been at the top of the priority list for manufacturers. Now that quadrotors are increasingly common and in use by non-professionals, the industry needs to take notice of the danger their products pose.
You mention that subjectively, the quadrotor was somewhat more difficult to pilot during testing. Can you elaborate on the potential downsides of the system?
We modified an off-the-shelf commercial system to show that retrofitting into existing products was straight-forward. In this case, we had to use slightly smaller rotors to get more hoop clearance, and since we could not return the flight controller to compensate for the change in size, it becomes more twitchy. This was a matter of a tuning problem due to a closed-source platform. The actual system itself has very few drawbacks—only a slight increase in weight and cost over an unsafe aircraft and the need to reset the safety system after a crash.
How much of a risk is there of one of the hoops being slowed enough to trigger the system in-flight by a non-impact? Is in-flight recovery possible?
After our initial tuning, a non-commanded triggering has not happened during our test flights. We can adjust how sensitive the trigger action is. We foresee that an aircraft with six or more rotors can employ a smart safety scheme where one rotor will stop in flight, while redundant rotors can keep providing thrust. Once the hoop detects that it is clear again, it can restart the stopped rotor to continue flying.
What do you think will be the biggest obstacle to adoption of this system in commercial products?
I see the biggest obstacle to commercializing this technology is that big manufacturers will not be prepared to accept the extra cost of making their aircraft safe. Even though the system is very cheap, it requires them to do more than the most basic implementation of their product, and that could have cut into their profit margins. I think savvy companies will realise that mitigating their liability for producing a dangerous product used by the general public is well worth a small hit to the bottom line, and potentially something that many people would pay an extra $20 for.
Have you (personally) tested the system?
I have tested it using a weighted mechanical substitute, in place of a rotor, and it worked great. My fiancée would be very mad at me if I tried it with a rotor for real, no matter how much I tell them that it’s very safe! :)
What are you working on next?
I’m always working on something different! We are just finishing a project to build a quadrotor that has three times the endurance of a conventional design— a two kilo drone able to fly up to 90 minutes carrying a GoPro camera payload. We have new airspeed velocity sensors that allow drones to instantly reject gust disturbances, making them safer and more reliable to fly around buildings. And, of course, a few more secret projects that I can’t talk about yet!
We’ve posted about some of Paul’s work before, including this hybrid quadrotor helicopter from IROS 2013. His electromechanical never fails to impress, and we’re very much looking forward to hearing more about those secret projects.
“The Safety Rotor—An Electromechanical Rotor Safety System for Drones,” by Paul E. I. Pounds and Will Deer from the University of Queensland in Brisbane, Australia, was presented at ICRA 2018.
Evan Ackerman is a senior editor at IEEE Spectrum. Since 2007, he has written over 6,000 articles on robotics and technology. He has a degree in Martian geology and is excellent at playing bagpipes.