Key lessons from our pilot testing the effectiveness of drones for controlling locust swarms in Kenya
Written by Nigel Breyley and Kush Gadhia
The Frontier Technology Livestreaming Programme has recently funded a unique project in East Africa to look at the use of drones in combatting locusts. Ground spraying teams are often used and the FAO of the United Nations contracts for airborne spraying of locusts by both helicopters and fixed wing aircraft. Consequently, experimenting to see if drones would be a natural gap filler between these two solutions seemed a good use case in this important battle against a pest which has recently caused wide spread devastation to food resources across much of the North and particularly the East of Africa.
The project decided to trial the use of drones specifically in hard to reach areas, to see if this technology could complement those existing ground and airborne spraying methods in an environment where it has traditionally been hard to spray effectively.
How did it come about?
Tristan Eagling, FCDO observed that “when the locust swarm arrived in East Africa in 2019 I was struck by the fact the methods we were using to control them were more or less the same as they had been since the 1950s and wondered what else could we do apart from the major control strategy of using planes to spray chemicals on locusts.
I had always been sceptical of the use of drones in spraying, and hadn’t seriously considered using them. But, being at the Drone Forum in Kigali in 2020 and speaking to Kush from Astral Aerial I realised that over the last decade the technology has moved on a lot and that spraying drones have much larger capacities and crucially can be a lot cheaper. There were also anecdotal reports of people informally trying drones for spraying locusts, so it seemed like a good time to do a more rigorous project to test their potential.”
What did we do to get started?
FTL adopted the project, which had been championed by Tristan, and after a selection process and comparison of potential providers Astral Aerial of Kenya came out as the winners with a compelling proposal. The company had the right expertise, offered a competitive price and their enthusiasm was obvious. Mind you, it wasn’t an easy choice given the enthusiasm of some of the other bidders and their vision of what could be achieved if the pilot project proved successful. We asked Kush Gadhia, Business Development Manager of Kenyan company Astral Aerial to tell us more about the company and why they had considered using drones for this role.
“Before the desert locust disaster struck East Africa, we were heavily involved in implementing drones in Agriculture. We had just successfully partnered with The World Bank to carry out agricultural mapping for farmers. This was so as to provide high resolution digital maps to farmers to enable them to plan their land utilization activities and assess extent of damage to crops following events such as heavy rainfall, as well as take measurements of areas covered.
In addition to that, we also provided crop health maps showing the health status of crops at a particular time, enabling early detection of pest/disease infestation and consequently, early mitigation of identified problems. This was carried out in counties such as Narok, Kiambu and Machakos for over 5,000 farmers at the time. We have also continued to carry out mapping services for over 16,000 farmers in partnership with one of Kenya’s largest Telco companies’ mobile-based agricultural services platform”
What else was being done to combat locusts?
It was important right from the very start that whatever solution was trialled needed to be one which could be sustained across what is often a poor or under-resourced part of the world given that the locust swarms had spread across Sudan, Ethiopia and other countries as well as Kenya.
In Kenya a joint locust fighting force had been assembled using aircraft and ground control units. The FAO funded force consisted of over 20 airplanes and helicopters, over 200 ground units with 3,000 newly trained spotters and control operators. These collaborative teams were able to spray control chemicals in vast flat regions where access by air, driving or walking was possible. In other areas however, where the terrain was irregular and dangerous to drive to, a more flexible aerial sprayer was needed that could safely fly near complex features such as down valley escarpments as well as spray close to the roosting desert locust swarms.
What were we looking for in a drone?
The proposal from Astral Aerial was to use the DJI Agras T16 drone, but we wanted to know a little more about what other drones might be available. It was obvious to us, and from our investigations also obvious to others that drones could have a role in finding and monitoring locust swarms. The FAO already had experience with that but could a drone really carry a large enough insecticide payload to be worthwhile. We needed to consider factors such as the relatively short endurance of battery powered drones and some other technical limitations such as the need for the pilot to visually monitor the flightpath and avoid obstacles — like valley sides in those hard to reach areas. Also among those considerations was how much ground support would be needed to get the drone, the chemicals and the people to the areas in the first place. Unlike a manned aircraft we could not afford to have the short endurance of the drone use up flight time flying in transit to the spraying area.
Astral Aerial had thought through the considerations too. “The fact that drones can achieve a more flexible flight course was a huge contributor for selecting our use case. Aircraft sprayers spraying over hundreds of square kilometres over roosting swarms, youth teams suited in knapsacks for ground control, vehicle and tractor mounted sprayers were already providing much-needed additional ground control support to tackle the swarms. Not to mention the supporting teams such as desert locust trackers, regional coordinators, desert locust tracking technology developers etc.
This obviously resulted in significant economic and operational implications to pull it off. Yet, we could also see that the planes were only economically viable when spraying over areas greater than 10 square kilometres. Knapsack sprayers could not spray on top of high trees, and tractors couldn’t pass impassable roads. Drone sprayers were a matching fit to provide the additional coverage to supplement our fellow desert locust responders”.
As part of our research we looked at what other users were doing with spray drones and one website struck a chord. Dronespray.com stated the following benefits of Smart Agriculture Spraying Drones on their website along with some useful figures:
The benefits, efficiency and cost reduction of this technology in agriculture are clear and proven. AG spraying drones can easily replace or improve on traditional pesticide, herbicide or fertilizer land-based sprayers.
Drone Sprayers can save up to 90% of water and 40% of pesticide use. Increased agriculture spraying efficiency: AG spraying drones can cover & spray up 40–60 acres per-day with a single drone operator
And then a final comment which aligned with our planned use case which recommended why drones were useful, so it looked like we were thinking along the right lines.
Easy access to spray hard-to-reach, elevated inclines such as hills or muddy areas and fields where traditional land-based vehicles cannot access.
What was available?
We did a desktop investigation to look at what drone solutions were available and our conclusions matched Astral Aerial’s in terms of affordability and utility. It didn’t mean that there were no other options available, just that none of them offered the same combination of advantages of the DJI Agras T16. DJI is a Chinese company which has become the world leader in terms of the number of drones sold worldwide, with options for tasks such as heavy lift, remote sensing, building inspection and of course agriculture. The T16 advertised the ability to spray up to 10 hectares per hour Agras T16 — DJI but we knew that there would be many other factors about spraying that still needed to be taken into account such as spray rate and operating height.
In much of Europe drone spraying for agriculture is not permitted due to concerns about drift of the chemicals. The reality is that the chemical drifts with the wind once it leaves the aircraft’s spray nozzles and often doesn’t land where you want it too, unless you are very low to the ground. Crop spraying using manned aircraft is big business in the United States though, but has a poor accident rate with many pilots dying each year due to the challenging nature of flying so close to the ground. That issue was one of the considerations which drove Pyka, an uncrewed aircraft manufacturer in California to develop their Pelican drone.
The Pelican looked good for our project from the perspective of the amount of chemical which it could carry and the coverage that offered in a single flight. However, the aircraft would also need to be towed by road between locations and required a take-off and landing strip. It also meant that there would be a need to charge larger batteries in the field and a need for more logistics support such as larger volumes of water for the spray mixing. In the confined areas we wanted to access it was unlikely to be as manoeuvrable as a multirotor drone with an ability to hover. Price too had to be another factor. Although the Pelican was impressive and the price very competitive compared to a manned aircraft, it was too expensive for a short trial and unlikely to be affordable in the longer term for the regions affected by the locusts.
Japan has a long established industry using aerial spray drones for agriculture, usually Kawasaki or Yamaha RMax helicopters, so these were considered along with multirotor drones from Quaternium of Spain and from XAG in China such as the XAG P40 Agricultural Drone with a 20 litre tank.
Most of these options had a very similar performance to the DJI Agras T16 in terms of the payload which could be carried, although endurance varied and of course price. What we needed to do was look at spraying time available and spray rates versus endurance. We tried to consider a metric which would allow us to assess cost per hectare sprayed, or number of hectares sprayed per hour etc. so get some indication of comparable costs and efficiencies with manned aircraft and ground based spray teams.
What did Astral Aerial decide?
We asked Kush about how Astral Aerial had made their choice and what they had considered:
“When deciding on the correct drone we had to factor a sprayer drone that would not only work for desert locust control but also for commercial farm spraying as that entails our day-to-day business. Through lots of research, we decided to go for the DJI Agras T16. We decided this as the team was already familiar with the flying DJI drones. We did evaluate several systems particularly the XAG drones but the deciding factor was really the ease of use. DJI drones’ are also readily available in the region making spares easy to come by.
There were some unique systems that could carry a much larger amount of chemical; upwards of 200 litres. These systems were unfortunately not really ready for commercial use and require ground infrastructure such as a small runway, which is hard to come by in some regions. These larger drones are also significantly more expensive and it was unlikely the authorities would approve a drone of that size as the drone industry was in its infancy in Kenya.
A hybrid option capable of staying in the air for over an hour proved to be quite interesting as the current use of batteries only allowed for around 12 mins of flight. Unfortunately, this hybrid system came in at around four times the price as the DJI, which was definitely a contributing factor to why we did not select that system.
Overall the DJI Agras T16 proved the best drone to initiate the testing of sprayer drones against locusts due its price and therefor lowered operating costs, plus access to spares and ease of use”.
Who helped?
We had the choice of aircraft, but of course that is only half the story. We had already considered the need for a metric measuring cost and coverage, but most importantly we needed to know if the spray method was actually efficient. Would it deliver the right dose or concentration of chemical required to kill the locusts and how could we optimise the spray mechanism to ensure that. In order to discover how effective the spray mechanism was Astral Aerial did a lot of in-house work alongside research that called on the expertise of Paul Miller from Silsoe Spray Applications Unit based in the UK: www.ssau.co.uk
Kush told us “We received guidance on the targeted spray rate which was a litre per minute. Initially the nozzles were not spinning at the correct speeds and we simply could not understand why. Paul managed to draw up a diagram and we quickly realized what the issue was. We added a potentiometer and by using this we were able to ensure the nozzles were spinning at the highest RPM”.
CABI www.cabi.org helped Astral Aerial to understand the desert locust behaviour to determine the optimum time of day for spraying that matched their roosting periods, as well as identification of the hoppers which is the stage of the locust life cycle where they appear most vulnerable to spraying. Most importantly, CABI had developed the ‘green muscle’ desert locust control chemical which we wanted to evaluate for drone spraying. To ensure that the right concentration of chemical was being used, invaluable support was provided by Ivan Rwomushana , who developed experiments to determine the optimum spraying heights as well as drone spraying modes to achieve maximum swath widths in terms of ground coverage. To measure this, sheets of paper were lined up on the ground; the drone flew over the papers releasing the spray. A mixture of water soluble ink and water was used as a spray in these experiments and the colour was easily visible on the papers. As a result, the Astral Aerial team was able to get a sense of the swath that was being achieved using ULV techniques. ULV is an abbreviation for Ultra Low Volume which allows more effective dispersal of chemicals Ultra-low volume — Wikipedia. In addition to these ink based experiments, spraying was conducted using 500 grams of Metarhizium mixed with 1 litre of diesel to spray live locusts in cages arranged systematically across a hectare.
Julie Makena looked further into the results of this work in her post: ULV drone spraying in the real-world environment | by Julie Makena | Frontier Technologies Hub | Medium
But what about Theory versus Practice?
The first field flights undertaken before the more scientifically based tests described above had required large amounts of water to be taken by road vehicle between spraying locations. On occasions where the spray areas were distant from the roads the team had to hand carry this water which of course can be an arduous and tiring process. The amount of water required for use in the field was also a major factor in the move to ULV spraying, a technique which uses diesel instead and which requires much smaller quantities of liquid to be transported.
To achieve the desired standard spray rate, engineering modifications were needed on the drone as well as the ULV atomizers. These include the drone’s chemical supply and regulation system, the power supply, nozzle positioning and orientation.
Pimping up your drone.
ULV sprayers can be fitted on drones relatively easily, but it required specialist technical expertise from the Astral Aerial engineering team to modify the spray rate and to create the unique positioning of the spray nozzles that would ensure effective spraying.
Important elements of this task included using a secondary battery to power the three nozzles, which need to spin at the optimum speed to ensure that the atomization was correct to target locusts. The drone also needed to remain as light as possible. Astral Aerial’s team took an innovative approach to solving that and headed off to a nearby scrap yard. They came back with some light aluminum window channel to create and extended boom to hold the nozzles. Kush told us that they managed to reroute a single DJI pump into the nozzles and the remaining were channeled to the tank to circulate the mixture, which avoided the extra weight of additional pumps.
The installation work kept Kush’s team busy for a day fitting the three ULV heads which they bought at about £350 each. The other purchases and installation work required a new 20,000 mAh lithium battery at about £400 to power the system and some additional piping at just under £75.
And the results were…?
The field trials had been conducted as well as more stringent controlled environment that allowed more scientifically measured results. So the question to be answered now was how successful and effective Astral Aerial felt the drone spraying of locusts had been. As part of that bigger question though and as part of the evaluation of this pilot project we wanted to know first what problems Astral Aerial had encountered operating in field conditions.
“For an initial field deployment there was a huge learning curve for us both practically and with regard to the drone. We made all kinds of plans and contingencies to ensure we had everything we needed, checklists, spare parts like propellers, dry runs etc.
Another major challenge was on take-off and landing space. As we did not have enough space within the bushy valleys to fit the large spraying drone, we had to carry out make shift landing techniques where the drone landed on one set of its landing gear, then have it pivot on this landing gear before it was caught by our field drone pilots.
Other significant challenges included hand carrying generator, batteries, fuel etc. over long distances as the roosting swarms were located far from the access roads. Generally, the drone performed impressively with minimal malfunctions or need for major repairs.
The major challenge we faced was on one of the least expected components, that is, the spray nozzles. The spray chemical was quite dense and left chemical residue on the nozzles, which led to clogging. But our quick fix for this was soaking the nozzles in water overnight.
We also identified the drone if not cleaned the night before would not hold altitude. We soon identified that dried chemical that would block the air pressure sensing holes to the barometer (altimeter). This was a challenge as it took some time to identify the issue but once we did we ensured the system was cleaned every night to prevent the issue happening again.
No drone operation of this challenging nature ever goes smoothly and while deployed in Ethiopia, the team suffered a crash with one of their drones. The Electronic Speed Controller (ESC) is a device which regulates the amount of electrical power sent to each set of rotor blades. This can therefore vary the amount of lift or thrust generated by each of the four sets of rotor blades. The ESC suffered a fault which occurred as the pilot was flying from one tree to another to spray. Kush told us “The drone somersaulted in the air before it crashed into the ground carrying nearly 13 litres of chemical. It happened in an instant. The pilot had no indication of this issue, nor was he able to perform a manoeuvre to stop the crash”. Fortunately no one was hurt, but the drone was damaged beyond repair. Although there was data available to the manufacturer about the cause, there was a lot of discussion with DJI about repair or replacement and eventually an insurance claim was submitted.
The metric for comparing cost for coverage and effectiveness sounds simple in practice but making comparisons proved complicated. We knew that the project had shown the use of drones to be an effective method of combatting locusts in these hard to reach locations. We knew the cost of putting a team into the field for a day (which naturally is commercially sensitive) and Astral Aerial and the scientists had shown that it was possible to spray about 5 hectares in a single flight. However, to compare the cost of drone airborne time with the cost per minute or hour of a manned aircraft isn’t straightforward nor does it offer a figure of much utility. Likewise coverage per flight or per hour couldn’t be read directly across between drones and manned aircraft. We could show that a drone team was more effective in terms of manpower than the use of ground spraying teams. But, what the project most certainly did achieve was to show at an affordable price that drones could extend the fight against locusts into those hard to reach areas which had previously gone untreated.
The Locust Lessons Learned?
Overall this FTL pilot project has been a success with many lessons learned by the programme and by Astral Aerial. So what are Astral Aerial’s five Top Tips and lessons learned? Once again we asked Kush, who has now become the CEO of Astral Aerial Solutions to manage the company’s expansion into other drone operations:
1. Thorough operations planning was essential, given the scale of the operation. That meant the use of technology like Elocust 3m for tracking and area size estimation etc.
2. Coordinating with other response teams was invaluable as they provided guidance on desert locust tracking.
3. Building valuable partnerships such as the one with CABI ensured our efforts were complimented by sound scientific methods and insights.
4. It would have been far easier if the ULV system had been available turnkey from the manufacturer. We had to spend a lot of time to ensure that it worked correctly.
5. Fly carefully — the increased width of the drone due to the long boom holding the ULV nozzles made the drone slightly less maneuverable in tight spaces, especially when taking off and landing.
