In 1870, agriculture was a source of employment for about half the U.S. workforce; now just 2% of the population produces food for a living, according to government figures. As large machines have increasingly done the work of ploughing and harvesting crops and tending to livestock, farms have become ever more human-free. And what of the future? They may provide barely any jobs at all.
The new age of farm machines include weeding robots, self-driving tractors, and crop-spraying drones. The drivers: higher productivity, an aging human workforce (principal farmers are 58 years old on average), and the cost of employing humans (California's minimum wage increases could lead to an automation wave, according to some reports).
"If you take a very long view [of] the world of agriculture, you can see that employment has come down as a total share of the world population, while the amount of food produced has gone up. Productivity has dramatically increased and we see robots as the next step in enabling that," says Khasha Ghaffarzadeh, lead author of a new report about agricultural automation.
Dairy milking bots are by far the most common forms of ag-automation so far. And, according to projections in "Agricultural Robots and Drones 2016-2026: Technologies, Markets, and Players" from U.K. research firm IDTechEx, they will still make up a lion's share of the market in a decade's time. Other technologies will increasingly come on stream, though, including "scouts" that check on the health of plants, auto-steer tractors, and data-mapping drones.
Ghaffarzadeh sees drones and robots hastening a move to precision farming, where crops are managed in small batches, or even individually, instead of as whole fields. That, in turn, could reduce indiscriminate spraying with herbicides and pesticides, and the waste of costly inputs, like artificial fertilizers.
Within a decade, automated combine harvesters could run pre-set routes, allowing farmers to get on with less repetitive activities (assuming these machines can be made safe for any humans who might still be in the way). But some parts of the farm may be beyond the capacity even of computer vision and algorithmic techniques. Soft fruits may still need human-plucking, for instance.
"If you think about the canopy of a tree, there are many fruits hidden away and the robot cannot see them. The process of taking an image, identifying the fruits, and planning the path of the robotic arm then gently but quickly picking the fruit is slow, expensive, complicated and hard to repeat," Ghaffarzadeh says.
He adds: "It may be that machines are designed to aid the human and reduce some of their time-wasting activities, like moving the [collection] bins and ladders around. That's easier to mechanize than the actual act of picking."
As with other industries, automation in agriculture is likely to be a mixed bag socially-speaking. On the one hand, we may get more and cheaper food, delivered more safely, with fewer nasty chemicals attached. On the other, farms might become increasingly industrialized, mono-cultured, and devoid of humans and human values.