Harvesting Mechanization
By Peter Keen
The future of both coffee and tea is one of accelerating the mechanization of harvesting. The economics of the industries and the demographics of the available labor force make that a certainty.
In coffee, Colombia is a good case in point. Its coffee production doubled between 2012 and 2017. The average age of plantation owners increased, and the number of young workers dropped. Farmers had to leave portions of their smallholdings unharvested. The National Coffee Federation (FNC), which represents 550,000 growers, estimated in 2015 that 20-40% more workers were needed to ensure quality harvesting of ripe cherries.
Where tea is concerned, we can look to India, where labor is half of production costs, even with wages that are under $2 a day. Kenya is marked by a great divide: industrial tea estates versus smallholder farmers. A strike of over 10,000 farmers in 2016 was over the refusal of large tea farms to implement a 30% wage increase. Other strikes over the past decade explicitly oppose mechanization as job killers. The view of the large growers is insistent: “Today without investing in mechanization, 36% of the tea fields in Kenya would be nonviable under hand-plucking.”
There is no plausible scenario that resolves the many labor dilemmas facing growers, policymakers, smallholders, and seasonal and local communities that are mutually irreconcilable. The machines are coming, and the jobs are going.
The equipment
In general, coffee and tea lag behind other crops, largely because of their terrain and fragmented scale of operation. There are four broad categories of harvesting equipment, with many variations in design, functionality, cost, and suitability to particular terrains, climate, soil, etc.: (1) portable, hand-held cutters and pickers; (2) larger, small-team devices that increase the speed and volume of crop harvesting; (3) tractor-like vehicles; and (4) an emerging class of vehicle integrated into a comprehensive communications-data-display infrastructure that adds computer artificial intelligence (AI) to machine muscle.
Terrain strongly constrains deployment of the options. Large elevated tractors can sweep along the tops of bushes cutting quickly and work well on large, flat fields where plants are organized in tidy, straight rows. Maneuvering a soil-disturbing massive ride-on harvester up, say, the heights, twists, and undulations of Darjeeling in the Himalayas is an obvious impracticality. But it’s perfectly simple for the massive fields of Argentina that produce half of US commodity tea imports; here, low costs and high volume, not quality and premiumization are the priority.
The capital cost is most affordable by the large grain, sugar corn, rice, and vegetable producers of developed nations but a near impossible stretch for the millions of smallholders whose farms may be little larger than a football field. For providers of mechanized harvesting tools, the large farms and high-value/volume crops are the natural market in terms of finances, expertise, and impact.
To a large degree, one of the main impacts is commoditization. The foundation of harvesting of the world’s two leading beverage crops has been the combination of the human eye and nimble wrist. The very term tea “plucking” of “two leaves and a bud” signals the precision and selectivity as does “handpicking” of only the ripe coffee cherries. These remain emblematic of the very best premium and specialty crops. Most of these are grown on mountain slopes at high elevations and varied terrain and soil. They are termed “specialty” and “premium” and cannot afford to be commoditized. Mechanization is not a solution to their business needs, which rest on branding, customer relationship-building, and impeccable quality assurance.
Mechanization has historically meant compromises in quality. For coffee, these include getting the maximum yield of ripe cherries minimally contaminated by damaged or green fruit. For tea, manual plucking of two leaves and a bud is far superior to the mulching of the bush. Some of the quality loss reflects lack of training, inexperience in adapting field layout and management, and other factors that can be remedied. Many experts believe that this will happen through a process of evolution in technology, services, and learning.
Here is a snapshot summary of today’s tools for harvesting coffee and tea.
Hand-held devices
These are inexpensive equivalents of hedge trimmers. They are simple cutters that speed up tea harvesting, generally with far less selectivity – broken pieces of leaf, twigs, tougher, lower leaf, etc. It’s more of a shearing and grabbing than a plucking.
They also make coffee picking easier, again, with compromises in quality. Derricandose are used in Brazil – vibrating and rotating rods at the end of a large stick that shake the coffee cherries off the bush – and stripping machines that gather ripe and unripe cherries, to be washed and sorted through enzymatic fermentation of water or friction washing and other methods. There’s no selectivity and the inclusion of unripe harvest makes for waste and added complexity in post-harvest processing.
The main feature of these handhelds is that they are light, with a battery- or gas-fuel-driven, typically two-stroke engine. The cutters will vary in width and strength. They may include built-in collection containers for the cropped leaf and cherry; in other instances, a second worker follows along to handle this. Many of these devices cost under $100. Figures vary widely on the productivity gains; many are marketing hyperbole. Overall, it seems that a halving of cost and up to 10 times increase in output per hour are reasonable targets.
Portable small team machines
These have become the main workhorse among Japan’s smallholder tea farmers. Common images are of an elderly couple handling the machine on their smallholder field. It’s an apt summary of the general issue of labor. Japan’s population is aging, and immigration is tightly restricted. Young people, by and large, don’t want to inherit the family farm and its burdens. Wages are very high.
A review of the widespread use of these two-person machines in Japan claims that they harvest up to 60 times per hour that of a single hand tea plucker but that the quality of the tea is reduced “drastically.” The analogy here is a lawnmower that is set to cut at a given height and can be positioned by its human operators. It is most efficient on flat terrain where the bushes are kept pruned and laid out in long rows. As for the mountain slopes and bumpy surfaces where many of the great teas are grown, the machines are of minimal benefit.
There are several features that help improve performance. For example, laser light is used to guide the cutter along a consistent height set to balance volume with leaf quality. Many smallholders use expensive hand plucking for the spring season harvesting, the best of the year when the buds and tips are at their sweetest. They switch to the machines for the later ones and in many instances are able to add an extra harvest through the improved speed and yield of mechanization. But the teas are not a match for the hand-plucked senchas, gyokuros, and matchas of the elite regions such as Uji and Kagoshima.
Customized tractors
When crop yield is the priority and quality a secondary consideration, tractor-style vehicles offer many advantages of scale and productivity – within substantial limitations of the types of terrain they can work in. The coffee harvesters are complex and can maximize the yield of standard coffee in a large flat field. They work top down, from the crown of the bush, shaking, stripping, collecting, and protecting the crop. Tea machines work just like a lawnmower, with the cutters elevated to the level that balances leaf volume and quality.
Several studies suggest that a quarter of an inch lower setting can increase broken leaf, twigs, and tougher fragments substantially. The major difference between skilled human plucking and basic mechanization is that the machine does not discriminate. That requires the addition of AI computer vision, machine learning, and neural modeling, all of which are progressing, but they are a long way off and it seems unlikely that robots will be harvesting on the mountains within a few years.
The main development is auto-steer driverless tractors. These use GPS receivers to keep rows straight and navigate a field’s terrain. Some have 100-foot-wide booms and include seeders and fertilizer systems that can be satellite-driven to within a few inches of accuracy.
The c.e.o. of Bear Flag Robotics, a firm that provides the sensors and control and communication mechanisms to add these capabilities to standard vehicles, is very explicit about the need for driverless equipment: “a crushing labor shortage, which drives up wages and labor mobility.”
Precision agriculture smart platforms
Defined and emerging in other areas of agriculture is the new generation of smart “decision platform” machines (an IBM term) fed by masses of data and satellite alerts, with machine learning AI tools for soil and disease diagnosis, weeding, and harvesting recommendations. These machines cost $150,000-400,000 or more.
This is where agricultural is moving on all fronts and in all regions. The tech giants are entering the market aggressively, with some tea and coffee applications but their efforts are mostly focus on building integrated platforms that provide telecommunications connectivity – including Microsoft’s tethered eye helium balloons a few hundred feet in the sky and use of television “white band” space in the signals that separate channels. IoT sensors gather field data on soil, irrigation, and weather.
Machine learning AI software has made immense progress in multispectral detection of diseased plants well before the farmer’s eyes and cameras can detect them. Drones are becoming a core and inexpensive flying farmhand team. John Deere is adding more and more capabilities to smart tractors, including the ability to spray individual weeds and plant seeds of exactly the right type in a given space.
It must be stressed that it will be a decade before this all comes together but that there will be continued if fragmented progress. Companies to watch include Brazil’s early adopters of drones such as O’Coffee, Strider, a coffee disease AI specialist and Falker, a software provider for coffee soil and irrigation management. In India, The Assam Company is committed to maximizing the use of mechanization across all areas of what is the largest tea producer in the world. John Deere and its subsidiary Blue River are the pace-setters in smart machines.
The social dimensions of mechanization
Mechanization comes with major impacts on the workers who remain in the fields. In Kenya, there has been strike after strike to block the adoption of machines that displace their jobs and their communities. Worker conditions and benefits have been worsening in many areas of the world. Poverty among Assam’s smallholders has been growing, along with UN accusations of slave trafficking, use of child labor, and failure to pay workers. The global retailers, distributors and brand leaders have strengthened their oversight and contract requirements, through Fair Trade and other consortia.
That said, the picture is bleak and the plight of workers in Latin America’s coffee and Africa’s tea farms compounded by the increase in droughts, climate change, and eroding margins that block investment, wage increases, and community support, including medical care and schools.
The irony is that wages are too low to attract and retain workers but too high to offset continuously eroding prices and increasing costs. The overall pattern in coffee and tea is captured by (1) India’s tea industry: input costs grew between 2015-20 by 8% annually but prices by just 1% and (2) by the routine estimate that labor amounts to 60-70% of coffee production costs, with harvesting comprising three-quarters of this. Both these figures can be switched between coffee and tea and across regions and countries. Mechanization is the new mainstream. Managing the journey is a business, technical, and social challenge.