In many recent discussions over the impact of automation on employment agriculture has often been cited. Between 1850 and 2015 the percentage of UK and US population employed in farming declined from 25% to 1.2% and 64% to 1.6% respectively. At the same time, production increased driven by labour productivity and mechanisation.
Today, farming in the developed world is highly automated with the cab of today’s advanced tractors resembling an aircraft cockpit. Amid the hype around autonomously driven cars in the future, the fact is that tractors have been autonomously ploughing, seeding, spraying and harvesting – positioned in a field accurate to about 1cm – for several years. The image above shows a recent autonomous tractor at work, one developed as part of the “Hands Free Hectare” project by agritech experts at Harper Adams University.
So, with this level of automation already, what more yield can we squeeze out of our farms?
The truth is that, whilst broadacre farming has undergone a transformation, there are still many areas of agriculture which are labour, land and resource intensive. In all areas of farming, there are continued opportunities to reduce input costs (and environmental impact) by tailoring the delivery of pesticide, herbicide, fertiliser and water to individual plants based on their growth state and condition.
Recent political moves to limit migration in the US and UK – which rely on seasonal migrant labour from Eastern Europe and Mexico respectively to harvest certain crops – have the potential to place extra pressure on the farming sector’s ability to maintain production without further automation.
The crops which remain labour intensive are not the broadfield arable crops or robust root vegetables. It’s delicate fruit and vegetables and those produced on perennial trees or shrubs with variable ripening times which still need human manual harvesting. These crops would be damaged by today’s mechanised harvesting approaches.
Crops in these categories include produce ranging from lettuce, tomatoes and mushrooms to asparagus, strawberries and grapes. And of course, peaches and lemons!
In 2016, it was estimated that cost of employment = (7.20 x 38hr/wk x 14wks) = £3830 = £210m/55k workers. Thus, 30% of output in soft fruit is consumed by labour input costs.
At minimum wage, the economics of this may be a challenge to fully replace people with robots. This could be changed by two things:
- If labour supply changes due to political moves to limit migration, the cost of labour will increase which would make robot automation feasible. In this context, we’d be talking about some of the complex robot ideas which can identify, select for ripeness, pick, quality assure, sort and store the crops. This is a challenging task for a robot, and whilst early machines are available (see Agrobot, Wall-Ye) these machines may take a few years to appear in significant numbers.
- An additional approach for the short term could be that we don’t attempt to replace all the workers, but to support them with robot helpers. Consider a robot “mule” which autonomously follows the human workers through the fields and carries the harvested crop. Once one “mule” is full, it can autonomously drive its way back across the fields to the store, only to be replaced immediately by a waiting empty mule, thereby allowing the harvesting to continue.
This example is a classic illustration of the challenge of making the economics of widespread robot deployment work. Should we aim for complex robotics which aims to remove humans altogether (and so is a bigger prize in cost saving) but is harder to achieve and may cost more, or should we aim for simple robotics which make humans more productive and could be deployed more economically?
It will be interesting to see the form of the next wave of robotic automation in agriculture. As is often the case, I suspect the answer will be a mixture of both approaches and whilst some individual attempts at agriculture robotics will undoubtedly turn out to be lemons, overall things are looking pretty peachy.