Ready for Takeoff: Drone-Based Sensing Takes Flight

MICHAEL EISENSTEIN, CONTRIBUTING EDITOR

Mines are inherently challenging places to work, but even within such an environment, certain tasks are considered particularly risky. In the aftermath of a blasting operation, for example, miners must assess the resulting opening in the rock and determine whether valuable ore can be recovered or whether additional excavation is needed.

The traditional method that miners have used to perform these critical functions puts them in a precarious position. “They were actually sticking a laser scanner on a pole into these cavities,” said Brandon Duick, CTO at Exyn Technologies. “It’s not the safest thing to have engineers or surveyors have to get that close to the space where there could potentially still be falling rock.”

Courtesy of Headwall.

Exyn offers an alternative: uncrewed aerial vehicles (UAVs), commonly known as drones, that use high-precision lidar sensors to autonomously navigate newly blasted openings, and generate detailed 3D maps that allow miners to plan their next steps from a safe distance.

Exyn is not the only company marketing such a solution, and a growing number of industries are finding opportunities to deploy sensor-laden drones in a similar fashion to acquire detailed environmental data that would otherwise be dangerous for human workers to obtain, or that pose logistical challenges.

“A lot of what we do is reducing risk for people,” said Adam Bilmes, cofounder and senior director of business development at Inspired Flight Technologies, citing the example of using drones to assess the condition of high-voltage power transmission lines. Some companies in this area, such as Quantum Systems, are providing drone-based sensing to assist soldiers on active battlefields. Others, such as AgEagle and Sentera, are developing airborne mapping solutions to evaluate crop health and identify sites of weed or pest infestation.

But tapping into these capabilities has its own difficulties. Even as more companies specializing in drone-based sensing enter the field, this technology remains expensive and requires a certain level of expertise that is still lacking in many industries. “We are not anywhere close to really broad adoption,” said David Blair, vice president and general manager of remote sensing at Headwall Photonics. “There’s still quite a gap that has to be closed.”

No mere toy

Drones have been popular playthings for hobbyists for some time. Their capabilities are evident in eye-catching aerial photographs, and their usage among nonindustrial consumers has shaped the narrative around the growth of the technology.

Over the past several years, drones have matured into useful tools for research, industrial, and other applications. On the one hand, regulators have established and implemented formal guidelines around how users can deploy UAVs. In 2016, the U.S. Federal Aviation Administration issued the set of rules known as Part 107 — Small Unarmed Aircraft Systems, which delineates how and where drones can be deployed and establishes certification requirements for commercial users. The European Union issued its own drone regulations and certification guidelines, which came into effect in 2020.

Drones themselves have simultaneously evolved, and in a considerable way. “A lot of folks building drones in their garage is really where this industry started,” said Bilmes — who added that Inspired Flight Technologies was founded in a similar fashion in 2017. But when the goal is to deliver reliable sensing capabilities to leading players in sectors such as agriculture, mining, oil and gas, or even defense, this approach simply will not fly. Today’s commercial UAVs are far more sophisticated and are precision-engineered systems. “We’re building aircraft that have to go through extraordinarily rigorous quality systems, and safety and reliability is everything,” Bilmes said.

In parallel, sensor developers have been turning their sophisticated instruments into compact, travel-ready packages that can operate stably in rigorous real-world conditions. Companies such as Headwall, founded in 2003, specialized initially in laboratory-scale instruments for hyperspectral imaging — developing tools for spectroscopic analyses of material and chemical properties of samples by scanning large swaths of the visible and infrared spectrum. “Around 10 years ago, we really started to recognize this inflection point where you have large spectrometers, computers, and data-processing systems all starting to miniaturize,” Blair said. Today, the company has downscaled commercial hyperspectral imaging sensors to a power-efficient 4-kg package roughly the size of a toaster.

Optical sensors have been shrunk to easily mount onto uncrewed aerial vehicles (UAVs) (top). Hyperspectral imaging sensors can now weigh <5 kg, offering users more options for deployment.

Scanning coral tanks onshore with uncrewed aerial vehicle (UAV)-mounted sensors allows scientists to gather valuable data in their mission to conserve and restore oceanic species along the Florida coast (bottom). Courtesy of Headwall.

Scaling down is only half the battle. These airborne sensors must also withstand a broad range of temperature and humidity levels, rapid movements, and vibrations in flight while still delivering quality results. Duick pointed out that even though miniaturized lidar sensors are already a common feature on cars to enable driver assistance and autonomy, UAVs are exposed to more challenging operating conditions.

“We try to fly as smoothly as possible, but sometimes you’ll have decent-size accelerations and angular rotations,” Duick said. “If we’re talking about millimeter accuracy, we have to impose very strict deformation requirements on all the parts.”

Many points of view

Despite these engineering challenges, would-be drone users can now choose from a wide range of integrated sensing modalities, depending on the requirements of their application. Optical photography and lidar sensors are among the most mature approaches, and both can be used to generate data-rich 3D maps of complex environments in settings ranging from construction and civil engineering projects to environmental sustainability efforts.

Photogrammetry is a more affordable option than lidar but may not generate the same level of fine detail. Also, photogrammetry is typically best suited for brightly lit or outdoor environments. For example, UAV manufacturer Quantum Systems recently collaborated with conservationists to deploy drones equipped with high-resolution cameras to find and classify endangered animal species, such as black rhinos in Namibia’s Kuzikus Wildlife Reserve. Lidar-based scanning may be a better solution in situations in which more resolution is needed — Exyn’s scanners can achieve 1-cm resolution — and is also generally considered the best solution for indoor or other low-light environments. “In a mining space with no light and lots of dust, the only way to really get the geometric information that those customers are after is from lidar,” Duick said. Exyn’s lidar systems have also found use in outdoor settings such as forest canopies, with the goal of assessing tree growth and health.

Lidar scanning can provide detailed 3D images in low-light environments — a powerful asset for navigating dark and dust-filled underground spaces, such as mines (top). Miners can assess the structural integrity of underground areas using drones while keeping workers at a safe distance (bottom). Courtesy of Exyn Technologies.

Spectral analysis sensors are also now gaining momentum for more specialized analytical applications. These include hyperspectral sensors, which can perform spectroscopic analyses across a broad gamut of wavelengths from the visible into the mid-infrared. As a lower-cost alternative to hyperspectral analysis, users may opt for multispectral sensors. These sensors can be configured to selectively analyze a more limited number of defined wavelengths, typically on the order of four to six, within the visible to near-infrared range.

Spectral analysis systems are powerful for use cases in which the goal is to extract deeper insights about chemical conditions in a given environment. For example, Bilmes said that Inspired Flight customers have used drone-based multispectral analysis to detect leaks in oil and gas pipelines. Hyperspectral sensing is more expensive to deploy, but it adds value in terms of its versatility, according to Blair, which can give it a potential edge in settings such as agricultural research or geological exploration. “A single multispectral few-band sensor can’t adapt if the environment changes,” he said. “Elements that show up and decline and transform are areas where we would rather focus.” These can include processes such as crop disease onset, changes in ecosystem(s) that are occurring over time, and water quality assessment.

Additional specialized instruments that are carving out a niche in the drone-based sensing market include UAV-borne infrared-sensitive thermal cameras. These devices are proving to be popular in military and public safety contexts as well as wildlife monitoring efforts. In these cases, the goal is to unobtrusively collect information about people or animals on the ground. The use of aerial magnetometric analysis is also expanding for geological surveys, including exploration for mineral wealth and the discovery and profiling of archaeological sites.

Some turbulence ahead

Many other considerations go into launching a UAV-based surveying or sensing program. Among them is deciding which of the two major categories of drones — fixed-wing and multi-rotor — should be used. Each has distinct advantages and limitations.

Fixed-wing drones, which resemble airplanes or gliders, can generally cover greater distances and offer superior stability to multi-rotor drones. But this class of drones also tends to be more expensive. Multi-rotor drones are more maneuverable and can hover in place for extended analysis of a site of interest, but they also travel more slowly and capture less data per outing. Some drone systems blur these lines; for example, Wingtra’s WingtraOne UAV uses rotors to achieve vertical takeoff and landing, making it easier to launch and recover while also achieving a balance of maneuverability and range.

Sensor-mounted drones enable users to gather critical data in dangerous or challenging environments. Courtesy of Norsk Elektro Optikk.

Drone-makers are also pushing toward higher levels of autonomy. Most UAVs are compatible with waypoint-based navigation, where the user predesignates a course on an already-mapped area and then allows the drone to go about its work. Exyn’s recently launched Nexys system offers a solution that uses sensor data to independently navigate its way through uncharted territory with minimal user intervention. “In general, if the lidar can see it, then our system and our algorithms will safely avoid it or scan around it as best as possible,” Duick said. These capabilities are particularly valuable in the context of unexpected events, such as a mining tunnel collapse, which can dramatically alter the site landscape relative to existing maps, he said.

On the sensor side, many sensor manufacturers strive to deliver solutions that are relatively agnostic in terms of drone compatibility, which gives users the flexibility to obtain a sensor-UAV pairing that best aligns with their needs as determined by application. “We actually don’t interface with the drone at all other than a mechanical interface and power line,” Blair said. “That way we can easily port from system to system.”

Nevertheless, quality concerns do come into play. Blair said that Headwall prefers to partner with established companies with robust customer support processes, for example. There are also geopolitical considerations. “Over the past few decades, we’ve really relinquished a lot of our capabilities overseas to China,” Bilmes said. Chinese company DJI, he said, manufactures today’s most popular consumer and commercial drone models. The U.S. Department of Defense has added DJI to a list of companies affiliated with the Chinese military, rendering it ineligible for many government contracts. But this has also triggered broader concerns around the commercial use of DJI technology as well as a lawsuit in response from the drone-maker regarding this designation and the resulting loss of business.

But Bilmes also pointed out that it has proved to be harder for American-made drones to compete in price. And indeed, the cost of entry can be high. For example, Inspired Flight’s commercial drones cost between $20,000 and $30,000 even before incorporating the sensor component, where sophisticated instruments also carry price tags of tens of thousands of dollars. This has limited the reach of drone-based sensing in some areas, most notably agriculture, in which most farmers lack the deep pockets of those in the mining or fossil fuel industries.

Not included in the cost is the need for operator experience. Bilmes stresses the importance of having a specialist within the organization who “owns” any drone-based surveying program and is well-versed in the safety, technical, and regulatory considerations associated with UAV use. “Somebody with good operations experience or a military background tends to be a very nice dovetail into this, or somebody that comes from an aviation background of any sort,” he said.

Automatic for the people

Obstacles such as these, when it comes to drone usage, are not insurmountable. According to Bilmes, it is realistic to obtain Part 107 certification for commercial UAV operation with just a few weeks of training, and certain organizations already offer drone-based sensing as a service.

Blair believes that costs will decrease as the sector expands and more early movers demonstrate the utility of drone-based sensing. He said that it is not always essential to pay top dollar to get good value from a surveying effort. “Especially when you start in a research market, there’s always this drive toward the best — the best specs, the best performance, the best image, the best everything,” he said. “The question is: Does it work for the purpose you need?”

A growing number of consumer-tailored systems are also entering the market. In 2023, Sentera launched its Aerial WeedScout platform, in which drones generate detailed maps of weeds down to quarter-inch resolution within a given field. These maps can then be used to guide the precise application of herbicides. The company reported that this turnkey solution can reduce the use of such chemicals by up to 70%, without negatively affecting the overall field productivity.

These systems will deliver an even greater value as they become smarter and more autonomous, with the capacity to perform more onboard data processing and decision-making. This will generally simplify aerial sensing projects while also making each run more efficient and productive.

Blair believes this to be an evolutionary leap that will be essential to the field’s growth.

“Until we really build robust workflows that a technician in the field can really operate robustly, routinely, and essentially unsupervised, it will still be in the early stages of adoption,” he said.