It’s summer in the northern hemisphere, and with it comes cool fresh salads, delicious ripe fruit and favorites like corn and other fresh veggies on the grill. But, looking down at your picnic plate, could you tell which produce may be soon picked by non-human hands? Makers of a new class of robotic farm equipment (and some farmers) may be hoping you can’t, or won’t care, as they usher in the next generation of Silicon-based farm workers to spray, manage and harvest our favorite crops.
Driving the thrust toward robotic farming is a mixture of economics, technology, and, in no small part, politics. At the same time we, as consumers, are demanding high quality produce at reasonable prices, a cloud is growing over the nation’s agricultural production. Proposed immigration legislation getting batted around Congress could make a huge dent in the substantially illegal workforce that both “big” and “small” agriculture rely on to pick many crops.
One government estimate places the size of the illegal workforce in agriculture at just under 50% between 2007 and 2009, or roughly 1.5 million workers, according to the National Center for Farmworker Health. Labor costs make up 42% of variable costs for fruit and vegetable production in the US, according to the US Department of Agriculture’s Economic Research Service (ERS). Withdrawing this low-cost source of maintenance, picking and packing labor, much of which is paid below minimum wage, could fuel rising commodity costs for farm-grown food at a point when the US consumer is still very sensitive to food price increases.
Add to this an already shrinking immigrant pool from which to draw, eroded by Mexico’s improving economy for one, and farmers are increasingly in need of new, predictable, and price-stable forms of farm labor—all of which, a shiny new worker on wheels is promised to provide.
Luckily for them, robotics specialists have been increasingly happy to oblige with new creations that pick everything from lettuce to soft fruits to mushrooms to cotton, at a rate that their inventors promise can increase efficiency twenty-fold, drastically reducing the number of workers needed to cover the same acreage. Additionally, they aren’t just picking, but planting, thinning, weeding, and even spraying pesticides from unmanned aerial vehicles, the latter of which becomes easier when removing humans from the mix, avoiding fears of contamination and illness through exposure to these chemicals. Robots even open up possibilities of changing the shape of the farm itself, automatically managing indoor hydroponic farms, for example. (Some are even considering how robotic farming could be applied in space.)
Some of the big technology shifts that have accelerated the development of farm robots have been in the allied fields of machine learning and machine vision. Robotics researchers have focused, in part, in recent years on teaching their creations how to spot a ripe strawberry, for example, and then gently remove it, skipping the unripe ones along the way. Companies such as Vision Robotics are teaching multi-armed orange picking units to capture 3D images of orange trees, and the fruit hiding in them, to pass on to grabber arms, while Stanford-launched Blue River’s robots are learning to visually spot weeds and inject lethal doses of fertilizer into them.
Still, researchers admit farm beds and planting techniques will have to be altered to meet some of this technology half way, creating new costs in order to accommodate these new forms of labor. Additionally, the handling expertise that a knowledgeable picker gains, which tells him or her just how lightly to twist a pickable peach, wine grape, or hot pepper in order to collect it without bruising, will take more time to perfect. And, the knowledge gained by experienced pickers of a particular field or orchard’s condition aren’t captured either by machine labor.
The direct costs of shifting from human to robotic farm labor aren’t yet favorable enough to make the return on investment incredibly attractive—some first-generation industrial units cost upward of $1 million, far more than the workers they replace (and some workers will still be needed, a la Amazon, to “manage” the devices). What they do provide is a lack of unionization headaches, a stable workforce that isn’t susceptible to seasonality, reactionary legislators, or the vagaries of a militarized border.
Robotic farm labor represents an area where particularly populist politicians might find themselves in an uncomfortable position, decrying both illegal immigration and falling wages. As some have pointed out in the recent debate about technology and labor, automation is already eating into the wages of workers in industrialized markets, and turning farming over to much higher levels of automation is not, in the case of the US economy, going to free labor to pursue higher skilled work. It may in fact further erode the immigrant base that in past eras refreshed our culture, and even today helps prop up the economy itself.
There are elements of the agricultural economy that will undoubtedly benefit from robotic labor. In the best case, a complementary economy might emerge, exposing some farm workers to opportunities to manage and maintain higher levels of automation, and not simply live in an either/or world of robot vs human. A close look at the new innovations touted in automated agriculture shows that, while amazing advances are being made, and new efficiencies found, bringing these costly units up to a level of skill and insight that an experienced farm worker possesses requires forging a huge number of difficult algorithms.
The promise of even lower cost food and fewer legal headaches shouldn’t mask the complexity of developing and deploying technology that can do the job required. We will be getting a greater array of machine labor on the farm in the very near future, but, breathless headlines aside, it probably won’t be panacea that both engineers and politicians hope for.