It’s no secret that drones have proven themselves an essential tool in enterprise and military applications. They save time. They save money. And they generally tend to improve safety. But drone crashes are still a genuine concern — and one American drone company has a plan to mitigate them.
Colorado-based Black Swift Technologies has developed an algorithm that can provide early warning and diagnostics of potential critical system failures on drones.
The U.S. Air Force in March announced that it had awarded Black Swift Technologies a contract to develop software capable of leveraging that algorithm in order to predict drone system failures before they happen. For now, it’s a small, Phase I award of just $50,000, set to run over the next six months. If the Air Force likes the tech, Black Swift could receive an additional $750,000 in Phase II money, and potentially several million in matching money as part of a Phase III should that happen.
With the funding, Black Swift is set to develop software that would use what’s called “unsupervised machine learning for anomaly detection.” In a nutshell, Black Swift would be able to construct a virtual model of how an aircraft should behave across a wide range of missions and flight conditions — and then watch for instances that violate these models.
Drones have been important to the U.S. Air Force in providing a “low-cost, high-mission capable asset when compared to manned airplanes,” according to a study by the Center for Strategic and International Studies.
Among the biggest factors that can lead to drone crashes:
- pilot error
- mechanical failure
- electrical failure
And among the biggest reasons why those errors occur:
- Most drones lack onboard monitoring or systematic maintenance.
- Pilots rely on guides printed in owner’s manuals (assuming they even read the owner’s manual), which may be out-of-date or not as comprehensive as necessary.
- A lack of subsystem state information for detailed maintenance logs and schedules — something standard for manned aircraft.
- Critical components such as servos are often open-loop and unmonitored.
- Inconsistent maintenance schedules.
- Fewer redundant systems (like multiple engines).
- Drone pilots might not have anywhere near the hours of experience that manned pilots are required to prove in order to get certified.
That said, here’s how Black Swift’s system for detecting drone crashes in coordination with the U.S. Air Force would work:
- The software would connect with avionics data already collected by the Air Force. If no data is available, what’s called a Monitor Node (basically a tiny computer) would be installed on the drone to gather it in real-time.
2. From there, data would be sent over the web to a dashboard, providing a color-coded diagnostic rating of each aspect of the drone.
That dashboard would be looking for anomalies (e.g.: failed sensors, low battery, lost comms, failed servo motors, damaged propellers, severe weather conditions such as icing, etc.), while also tracking ongoing drone maintenance.
Black Swift Technologies has been around for about 10 years with an initial emphasis on building drones that could fly scientific payloads in complicated atmospheric environments, such as in high-altitudes, in extreme climates like the Arctic and deserts, or in strong turbulence. Most of its applications so far have been in atmospheric research missions in extreme conditions, including monitoring and assessing wildfires, volcanoes, tornadoes, and hurricanes, and its drones have been used by NASA and NOAA.
In that time, Black Swift has score a number of interesting government contracts, including a contract with NASA to design an aerial vehicle capable of conducting upper atmospheric observations of the planet Venus. NASA also once partnered with Black Swift Technologies to create a set of drones called the SuperSwift XT, that can be sent around volcanoes with sensors that can measure gas and atmospheric parameters, gathering data about particle size-frequency distribution, vertical ash concentration and levels of sulfur dioxide.
And just earlier this year, NOAA selected Black Swift to develop commercially viable technology enabling GPS-denied navigation of drones, which is essential to enabling long distance, beyond visual line of sight (BVLOS) flights.