Once solely within the domain of the military, drones are now critical aerial workers that perform a range of commercial surveillance tasks—from monitoring oil rigs to surveying agricultural fields.
The economic value of this flying workforce reached about $1 billion in the United States alone in 2017. By 2026, that figure is expected to grow to at least $31 billion.1
That growth means more drones buzzing the skies, collecting loads of data.
According to one estimate,2 a small drone fleet could produce 150 terabytes daily (that’s about equivalent to 75,000 hours of HD film3). And businesses will rely on new technologies to parse and store that data, says Vivien Heriard-Dubreuil, President of Microdrones, a drone systems maker that sells solutions globally.
Better data-processing capabilities, through the rise of edge computing and 5G, will help support the massive data caches created by surveillance drones in the near future—and give drones more autonomy at the hands of artificial intelligence.
We sat down with Dubreuil and his colleague Mike Hogan to talk about how smart drone technologies will help companies put the terabytes of data their surveying drones collect to work. (Responses have been lightly edited for clarity.)
Interview with Vivien Heriard-Dubreuil and Mike Hogan
When a business comes to Microdrones for surveying, what is the most important part of making sure the job is done well?
Heriard-Dubreuil: What matters most with drones are the applications—what data users want to collect.
For example, at Microdrones we have a special laser sensor designed to detect gas pipeline leaks. Before buying the system, customers typically challenge us to detect known leaks in a testing facility.
With the gas detector drone, our current setup can log up to 10 readings per second while scanning an area where a potential gas leak can be present. These readings are accompanied by navigation data—for example, altitude, longitude and latitude. The data from these sensors are recorded on the onboard SD cards.
What other types of data do the drones collect?
Heriard-Dubreuil: Some of the data is from cameras: Most of our drones are using low- and high-resolution cameras to produce pictures and videos. The data from these cameras can run 10 to 16 GB per flight. We use 64 to 256 GB capacity SD cards to store the data during the flight.
We also collect sensor data—for example, the Light Detection and Ranging (LiDAR) scanner is one of many sensors used in several of our product s.
It generates about 1 GB of data per flight, on average. Using an app, surveyors can track and control the drone using a tablet in the field.
Can data be transferred in real-time from the drone to the people on the ground?
Heriard-Dubreuil: Yes. Users have an option to use local LTE networks, but a lot of the projects take place in rural areas. If they are out of coverage, they need to use specific downlinks; most of our systems today are based on MHz frequencies.
5G networks are already available, and are likely to be widespread in the near future. How will that change the way your drones operate?
Hogan: Based on the current regulations, a lot of drones probably use point-to-point communications.
Like Viven mentioned, in a lot of the areas where people operate drones, there might not be the LTE or the cellular communications. This is always the biggest challenge because of the limitations in communications: The data stays on the drone.
When I look at 5G, I see a whole new opportunity to transfer data in real time, rather than waiting for the drone to land and then use the data.
It’s a paradigm shift on how we look at drones. If the limitation isn’t the bandwidth of getting the data, now we can think about how we can take advantage of that data in real time.
Does edge computing also play an important role in how drones operate in the future?
Hogan: It’s important if you want to do some kind of analysis on the data.
Today, processing the data on the ground is the safer thing to do because of the computing power that’s typically required, and because you start to take power and physical space on a drone. But with access to that low-latency data transfer, then you can effectively process in the sky.
With the ability to process and analyze data powered by 5G and edge computing, do you see that enabling artificial intelligence on the drones?
Hogan: In any application, we ask a question and then we want an answer. In order to get at that answer, we need to collect data, we need to process data, and we need to analyze data.
It’s funny, people think about drones and say, “Oh, now we have drones, it’s unmanned!” But the reality is, there are 15 people on the ground managing it. AI could start to take the human out of the loop.
AI, as we go forward, is bringing all the different systems together and starting to make the decisions for the drone.
This idea of unmanned drones is a cause of concern for a lot of people—especially when it comes to unmanned surveillance. Do you think that’s a major hurdle for the industry as 5G, edge computing, and AI make drones more effective?
Hogan: At least here in North America, surveillance isn’t done haphazardly. We have our rights and freedoms that are protected—whether it’s a drone, or if it’s somebody with a camera.
I look at drones as another tool, for whatever market you’re in. It could be police, firefighters, surveyors, people doing inspections, emergency responders. It can make these jobs safer.
For example, if you’re inspecting cell towers, you don’t want a human to have to climb because there’s a risk that someone could fall. There are all kind of applications where it’s positive.
Read More on the Data Powering Surveillance Systems:
- Deep Learning in Surveillance
- How Smart Video Surveillance is Changing Edge Architectures
- Durability and Surveillance in Edge Units
This content is produced by WIRED Brand Lab in collaboration with Western Digital Corporation.