As the hype fades and more growers test out drones on their own operations, what they really want from drone-focused technology is becoming clearer and raising the bar for startups in the field.
Ewan McFarlane, head of digital agronomy at Origin Enterprises in the UK, supervises 750 agronomists across western and eastern Europe and though his company has used drone services on a few occasions, it still does not own any drones.
Farmers using basic drone models, most commonly the Phantom 4 from dominant Chinese manufacturer DJI, report to McFarlane that the return of actionable information is not worth the cost of the labor to operate the machines.
To gain more relevance, either the data and analytics gleaned from drone imagery have to offer more value, or the drones need to be able to operate with less human interference.
According to the Commercial Drone Industry Trends Report from DroneDeply, a major provider of drone flight and mapping software operating across many industries, drone pilots with mapping capabilities charge an average of $168 per hour. Even if a current employee on a farm gains the skills to operate a drone, his or her time is then taken from some other task.
Join Us! Sign up for our next fund here.
And the cost of operating drones doesn’t go down on larger farms: an analysis by Ipsos Business Consulting on the cost of imagery gathered by drones compared to satellites and fixed-wing aircraft found that for farms larger than 25 hectares, drones are the most costly option, while on farms smaller than 25 hectares, they are the most cost-effective.
It follows then that the same report shows drone adoption in Japan, where the farms are smaller, the terrain is often more challenging, and drones are used more for spraying and seeding, at 70%. This compares to the US where farms are larger and drones are more commonly used to take imagery for precision agriculture applications and adoption is at 40%.
“There is a recognition that it really is too expensive to put a guy in a field,” said McFarlane, pointing to increase increase in autonomous drones. “Machines that can charge themselves on a base, go out and fly a mission, come back to base, and upload the data they’ve acquired without any human intervention: I think there’s a sense out there that that’s probably the direction of travel that’s going to yield a low enough cost of acquisition per image to scale the service.”
But autonomous flight is not enough to fully replace that need for human labor and the resulting cost, argues McFarlane. Many drones can integrate with software that allows for a degree of autopilot, but the battery charging, data transfer, and proper storage still requires a human intervention.
American Robotics (AR) is one startup working to remove that human altogether. The Boston-based startup calls its completely autonomous drone, Scout, the first “practical” drone system. Flights for Scout can be planned and scheduled from a computer. The drone itself, of AR’s own design, is kept in a box in the field that resembles a large beehive, where the drone charges when not in use. When it’s time for a mission, the top of the box slides open and the drone takes off.
AR CEO Reese Mozer said that his system, which is sold as a service with no upfront purchase of hardware, is cheaper than current alternatives when labor is factored in.
“When you take the human out of the loop and you handle the data processing in a more intelligent fashion, it provides some very interesting economic results. For the vast majority of farms our solution will be more much cheaper over the course of a year,” said Mozer. AR closed a $1.1 million seed round in March from angel investors and included Brain Robotics Capital, a fund focused on AI, robotics, and IoT.
A snag in the plan for AR is Federal Aviation Administration (FAA) regulations. Currently, in the US, drones must be flown in the line of sight of a human operator, which lessens the value of autonomy, especially on very large operations or noncontiguous farms. But Mozer told AgFunderNews that behind-the-scenes progress is being made on the issue and that new regulatory concessions may be on the way.
“The ultimate goal is to have the system operate continuously beyond the line of sight as a lot of farmers don’t have a continuous plot of land,” said Mozer. AR is targeting farms with at least 1,000 acres of broadacre crops and possibly less for high-priced specialty crops. Mozer said that there was no ceiling on the size of the farm that AR’s Scout could work for, but multiple drones may eventually become necessary.
What About Spraying?
Taking on autonomy from a different angle is Israel’s Skyx.
The traditional methods of pesticide application such as planes, tractors, and handheld sprayers, are inefficient in terms of both labor and cost. They also offer environmental risks such as pesticide drift and water contamination, as well as risk to human health for growers and workers in close, frequent contact with the chemicals, according to founder Eylon Sorek.
Skyx’s software allows the user to control multiple drones — what it calls a “swarm” — and direct them to complete a single mission. The software allows the user to build a fleet of drones gradually over time and choose the drones themselves, as Skyx intends to be compatible with any commercially available hardware. The company raised an undisclosed seed round from Israeli investors last month.
“The reason that we don’t see one person and one drone in the field today is because this person doesn’t offer enough return on investment. With a swarm, you can achieve a competitive price point with the exact same operation,” said Sorek. Though Skyx’s solution doesn’t take the human operator out of the equation, it does increase that human’s capacity to manage multiple drones— a sort of halfway mark between manual operation and full autonomy.
A limiting factor to this technology is the frequency with which spraying drones need to be serviced, pesticide payloads refilled, and batteries replaced. Sorek says that a reasonably-sized fleet for one operator to service is four to six drones, which can reasonably cover 40-60 hectare (100-150 acres) in one day — that’s appropriate for small and medium-sized operations, but not large row crops operations, according to Sorek.
Also working on a level of autonomy in spraying is Japan’s Nileworks. Using artificial intelligence, Nileworks claims its multi-copter drones are able to see the shape of a field and spray just 30cm above, thereby reducing drift. The company is also seeking to add diagnostic capabilities to the drones, enabling them to decide how much pesticide is necessary based on the appearance of the crop. Nileworks raised a ¥800 million ($7.12 million) financing round from Japanese investors in October.
The Tokyo-based company, founded in 2015, will target rice farmers in Japan when the product goes on sale in 2019 but says it will move on to other rice-producing countries in Asia in 2020. The company makes no claims about autonomous charging, and obviously, with any drone carrying chemicals to be sprayed, the payload will need to be refilled. But the “see and spray” capability that Nileworks purports to have could answer a common complaint of growers — that drones aren’t helpful enough with decision-making. Making spraying decisions independently would certainly meet that standard as long as the decisions are appropriate.
According to Ipsos, the agricultural industry climate in Japan is particularly conducive to the adoption of drones since the country has created an amenable regulatory environment and productivity is a real challenge.
In the early days of using drones to capture aerial imagery, just having an aerial image of your farmland added great value. The idea was that farmers could fly their fields as often as they wanted to pinpoint issues, such as irrigation leaks, leaf color variation, or pests like nematodes. Soon, however, this information was not enough and farmers complained they were getting less and less value out of the images. While the images could help them plan out their days better by highlighting where these issues were occurring, an argument was forming that by the time the imagery told them about certain issues, it was often too late to remedy the situation.
“A general comment that you get from a lot of farmers is that they feel that the information that they’re getting, by the time they acquire it, they feel that they’re staring at a rearview mirror but it’s not good for making future decisions,” said McFarlane.
Read more about the companies attempting to turn drone imagery into actionable intelligence here.