The last agricultural revolution turned out dumb machines to replace human muscle. Now they get a brain.
Consider the tomato.
Lord knows, David Slaughter has. Tomatoes, weeds, dirt clods. This is the man's life - seriously. Professor Slaughter and his comrades are plotting a little revo in the barn. Over the last few years, they have begun to hammer together the power of satellites and silicon, laser beams and machine vision, and they are bringing these tools to work in the big old industry formerly known as farming.
Of course, they don't necessarily call it farming anymore. How about "bioproduction"?
Sound boring? It is true that I sat, pen and paper in hand, while an earnest engineer here at the University of California at Davis leaned back in his chair, got that dreamy look on his face, and whispered, "What I'd really like to develop is a good mold sensor."
Well, who wouldn't?
Maybe the Artificial Vagina will get your attention. How about a Bug Accelerator that blasts magnetized, predatory lacewing larvae out of a charged nozzle gun? An expert software system for Manure Management? The Robotic Teat-Cup Attacher?
Imagine amber fields of grain filled with robo-tractors, operating without a driver, not a bibbed overall in site, guided by geostationary satellites and machine eyes, searching for ... alien aphid infestations. Attention undergrads: this field is ripe. Because at present, a lot of this stuff doesn't work.
But it will.
The inventors who are bringing these emerging technologies to the farm number in the thousands at dozens of the world's premier agriculture schools and research shops. Indeed, the last gathering of farming's digital tribe drew 600 participants at the University of Minnesota, where they do damn good work. With wheat. But life is too short to focus on grain alone. The curious pilgrim seeking one institution through which to view the future of farming in all its cornucopian splendor could do no better than visit the University of California at Davis, the sun-baked sprawling state school where digital farming got its start in the clunky 1960s.
For not only are the researchers at Davis some of the biggest brand names in the business, but their "problems" are so damn interesting. Sure, they've got wheat. But more deliciously difficult, they've got all the nuts, fruits, and vegetables appropriate to the California setting. Kiwi, oranges, almonds, grapes, peaches, tomatoes. Stuff that grows on vines, trees, in rows. They've got environmental constraints. Water shortages. And an enormous amount of money. And they've got David Slaughter.
Slaughter and his busy graduate student, Won Suk Lee, are down in the laboratory in the bowels of Bainer Hall, working on their latest version of the Digital Weed Destroyer. Think of a wagon, the size of a kid's go-cart, that will be pulled behind a tractor - at least for now (later it might cruise along, steered by satellites or laser guidance systems). Mounted aboard the wagon are two rather ordinary videocameras, the kind they use at the mall for shoplifter surveillance and to provide changing-room soft porn for security guards.
The forward camcorder steers the wagon. It takes 30 images a second. A frame-grabber picks a picture and transmits it to the computer, which analyzes the pattern (looking for baby tomato plants, known in technical jargon as "green blobs"), runs the mathematical algorithms to find a centerline, and so navigates the row, trundling along a path only inches wide as disks and blades chew the soil, whacking weeds that grow along the sides.
Now the cool part: The second camera is not staring ahead, but is looking down below at the row of tomatoes and the weeds growing among them out of reach of the disks and blades. The images flash to the computer, which has been loaded with pattern recognition software and is frantically trying to figure out what is the Red Gold of the Sacramento Valley and what is not. "This is the tricky part," Slaughter says.
At present, the typical large-scale tomato farmer must hire a crew of 40 men and women - mostly migrant workers from Mexico, and we will get into the social implications of this stuff later - to stalk the blazing hot fields with hoes to dispatch the weeds in the rows. "With a human, he knows immediately this is a weed, this is a tomato." A field hand from a small village in Mexico constitutes the best microprocessor on the market. "Totally smart," Slaughter says. "Processing color, texture, location, shape. One stroke of the hoe and the weed is gone."
The pattern recognition of the video-computer system is not so accurate. When a tomato is young, it possesses just two delicate, very hopeful seed leaves, bristling with little plant pubic hairs. But they do not always open in the same way on the same day. The weeds are essentially look-alike tomatoes. Slaughter needs his computer to immediately recognize "tomato" and give the all-clear that every "nontomato" in its path be terminated with extreme prejudice.
At the rear of the wagon is an array of nozzles, loaded with herbicide. When a weed is spotted, the computer gives the order: Death from above. And a precise squirt of liquid death is issued on just the weed. As Slaughter puts it: "Why spray dirt?"
At this point, Slaughter gets that dreamy look, too. "Of course, it doesn't have to be a chemical," he admits. The boys in the white coats are spinning other extermination scenarios. From the wagon's platform, the weeds could be zapped by a laser or electrocuted with an electron beam or blasted with a heat pulse from a high-powered magnifying glass that clicks on and off, powered by the sun, controlled by a shutter that winks open and shut. A personal favorite: they could also be decapitated with the Magic Switchblade, a robotic scalpel dancing around, slicing and dicing across a digitized grid. "There could be a piece of machinery that might be cheaper to use than chemicals," Slaughter suspects.
But then again, maybe not.
The Digital Weed Destroyer right now is only about 60 percent accurate. The problem is recognition. I am thinking, Whew, that is a stumper. How many PhD students is Slaughter gonna burn out on this one? Then the professor turns and points at a tiny tomato seedling, sitting in its black plastic cup, alone on the lab counter.
The only difference: this plant is purple.
Seems Professor John Yoder and the gene splicers across campus have developed a new strain. The implication is obvious: acres of purple tomato plants basking in the sun, as the Digital Weed Destroyer makes a much easier decision.
Kill anything green.
They are calling their movement "smart farming." Under its rubric is a long list of emerging technologies, some inexorably destined for the rubbish heap, some already online - including aquaculture, hydroponics, and, of course, chromosomal manipulation, which leads to the infamous "Flavr-Savr" DNA-spliced tomato and inevitably to a whole produce section filled with the genetic bastards.
On the digitized side, the hottest concept currently is "precision farming" or "farming by the foot," driven by the idea that if machines and computers can help a farmer apply just the exact amount of disincentive and encouragement, exactly where it is needed, it will not only save billions of dollars and jack up profits, but give the farmland and the surrounding streams and forests a much needed respite from the relentless dousing of nasty and wasteful fertilizers, herbicides, and pesticides.
This is not a trivial matter, or one of concern only to farmers. There are already fields and groundwater so poisoned by nitrate fertilizers in places like California's San Joaquin Valley that environmental activists have begun to use words like "moratorium."
If one likes to think of technologies as waves, smart farming is the third in agriculture - the first being the invention of farming somewhere in the Near East, when humans stopped hunting and gathering and started poking pointy sticks in the ground and planting seeds, which led to related phenomena like building cities and creating political and mercantile cultures, to say nothing of spinoffs like pyramids, subsidized art, slavery, and large-scale war.
The second wave was the move from beasts of burden and human labor to steam- and then gasoline-powered machines, the great Mechanization of Agriculture, supercharged by the Green Revolution of cross-bred crops.
This was the wave that sparked, among many other things, the mass migrations in post-World War II America of blacks from the rural South to urban North, the hollowing out of small towns across the Great Plains, the disintegration of a social structure to the point where lonely males who inherit the family spread now advertise in the city papers to turn up a potential wife.
Meantime, though, food has become so plentiful that governments in the Western world still pay their farmers not to plant, and famine has became more a political problem than a natural one.
The march toward third wave precision has the feeling of inevitability about it. In a natural world of tremendous variability, farmers are the ultimate control freaks. However much they want their fields to hum like factories, they are still plagued by the immensely complex realities of soil and rain and bugs. Yet in a competitive, post-NAFTA marketplace, these bioproducers make their money by the slimmest of margins. A few pennies more for each pound of almonds or bucket of milk is the difference between sending the kids to college or converting the old homestead into exurban housing developments with names like Green Acres.
"Farmers might not all be environmentalists," Slaughter says. "But they are all businesspeople."
Darwinian indeed.
You've got to respect a university that has a real dairy farm smack in the middle of campus. At four in the morning, the stars wink in the heavens and the air outside is cool and sweet, as Rod Claycomb, one of Professor Michael Delwiche's PhD students, directs me inside the Dairy Center and introduces me to Elsie and her 79 sisters as they eagerly queue for their predawn turn at the automated teat suckers.
As much as dairy farmers might like to think of their cows as mere milk machines, they remain stubbornly biological, therefore fickle and hard to predict. In a perfect world, a dairy cow would always be nursing and producing an Old Faithful of milk. After she gives birth, a dairy cow's milk peaks somewhere around 125 days postpartum and then declines.
"The farmer wants to keep her pregnant all the time," Claycomb explains, himself the son of a Pennsylvania dairy equipment dealer. "But the failure to catch estrus is the Number One problem in the dairy industry."
A cow operates on an approximate 21-day cycle, as opposed to 28 days for someone like my lovely wife. Twenty-four hours before a cow ovulates (the exact date and time is impossible to predict), her reproductive window quickly opens and then shuts down in a flash. Around the time of ovulation, a cow will "stand" to be mounted by a bull. Of course, most dairy cows have never seen a real bull in their lives (male dairy calves are called "veal scallopini").
Instead, cows are artificially inseminated by a manager or veterinarian whose job it is to put on a giant rubber glove that reaches to the elbow, fortify themselves with pleasant thoughts, and then ram a fist up the cow's posterior. Talk about a digital revolution. Once waaaay up there, the vet holds her cervix through the colon wall, keeps it steady, and inserts through ever more private bodily portals a pipette filled with US$5 worth of bull semen. How they get the sperm at the stud farms is where the Artificial Vagina comes in - it's an inflatable whoopie cushion filled with warm water and lathered with bovine KY jelly and a glass test tube hanging off one end.
But enough about him. What about her?
If all goes well, the cow is impregnated. But it does not always go well. Months can go by. And a $1,500 dairy cow that does not get pregnant when the farmer wants her to is soon converted into $400 worth of Hamburger Happy Meals.
The trick is knowing exactly when Elsie is in estrus. This is where "smart farming" comes in. Currently, a dairy farm might hire a bunch of minimum-wage "cow watchers" who stand around just to see if Elsie is in the mood. To oversimplify what I am sure is an art form, they scan for any sign that the female will be mounted. Females mount other females in the dairy industry. Don't ask why. It is none of our business.
This "cow watching" is faulty and expensive. Some farmers have begun to outfit their cows with pressure sensors that record any attempts to be mounted. This does not work with great accuracy, either.
And so we come back to Claycomb and Delwiche. They are developing a biosensor, one of the hot technologies in smart farming, which samples milk as it is being sucked still steamy-warm from the cow and, within 10 minutes, generates readings of the hormone progesterone.
"It's sort of like a home pregnancy test for cows," Claycomb says, fiddling with this Rube Goldberg apparatus filled with tubes, milk, microliter sample wells, diodes, and amplifiers. We look at the printouts of the cows being monitored and there it is: the wild ride of hormones, day to day, recorded at each milking, eerily pinpointing, within hours, estrus.
If nothing else, this is going to make the men and women in the big rubber gloves very happy. It is also going to make milk ever cheaper to produce. That new smell in the dairy parlor? It's not manure. It's money.
We're back in Bainer Hall with Michael Delwiche, discussing biosensors like his Cow Hormone Clocker that I saw early in the morning. We're talking cows, when suddenly Delwiche interrupts himself, digs a nice soft peach out of his lunch sack, and holds it aloft. Like many fruits and vegetables, the trick is not just growing the stuff, but "postharvest," the handling, grading, sorting, and shipping. Biosensors are a very enticing possibility.
Delwiche cradles the peach in his palm. "Look at it as an engineering challenge," he says, though I confess I do not like to think of things I put in my mouth that way. A farmer and his produce packers want a peach that arrives at the grocery store not too hard, not too soft, but firm, fresh, and ready to eat in a day or two. Delwiche calls it the "soft-hard problem." You have to love him for it. "A human at a grocery store can pick a melon. They can squeeze it, smell it, heft it," Delwiche says. "They know exactly what they want from the melon."
Delwiche searches around and shows me his Hand Fruit Impactor. It has an air cell, an accelerometer, some hardwiring, a bit of microprocessing. When the peach is presented, the Impactor bumps it with a mini-battering ram, a hopefully "nondestructive impact" that attempts to judge firmness. It is not yet a highly accurate device. The little old ladies squeezing all those melons at my local supermarket have them beat. And it gets worse. Peaches, like other vegetables and fruits, pose another problem.
Defects.
Like David Slaughter's Tomato Weed Destroyer, machine eyes are being tested to sort out the rejects. But a peach is not uniform and round. It is covered with natural landmarks - the stem cavity, which looks just like a black hole, a blossom end, and the suture, that lovely curving ridge that always reminds me of a nice butt rather than an engineering challenge. Even with two or three cameras and a rotating fruit conveyer belt, one can imagine the difficulty. At present, the sorting is still done best - and most often - by people.
Delwiche wants to tackle the problem not only with machine eyes that "see," but with biosensors that "sniff." It is all a question of identifying the right combination, the perfect perfume of near-ripeness, and then using a biosensor to whiff the volatiles and making a computer-assisted decision. "This is a 10-year, a 20-year problem," says Delwiche, which almost seems to please him.
The first farmers to eagerly embrace the concept of precision agriculture have been the Midwest agrarians raising immense fields of wheat, soy, and corn. There, a tractor is as likely as not to be outfitted with an antenna to slurp coordinates from the Global Positioning System. The GPS satellites not only tell Farmer Brown where he is in a field, within a few feet, but are being used to create ever-more accurate maps of his acreage.
A mapping firm, and there are now hundreds of them, goes into the field, takes a pinch of soil, and records myriad measurements - of salinity, acidity, nitrogen, moisture, temperature. These data can be combined with records from years past of infestations and production - how much each few feet of cropland yielded - to create maps with pretty colors. And so as the computer-tractor crawls along, it can help make decisions, such as dump more fertilizer here, spray less insecticide there. Information equals application. As UC Davis Professor Ken Giles put it: "There's no need for a thermonuclear strike."
The still-evolving concept goes like this: superaccurate GPS-guided maps, coupled with a field bristling with little biosensors, would instantaneously feed information to the onboard computers mounted in the tractor's air-conditioned cab. The instructions could be complex - a 100-acre field of cucumbers can generate a fire hose of data hundreds of pages long. Then again, the software could squeeze the flood into a few simple words.
"Water me."
"Feed me."
Or as I like to imagine: "AArrgggg! I got bugs all over me, spray me, quickly, hit the red button, you idiot!"
Bombs away, boys.
Professor Shrinivasa Upadhyaya drives me out to one of the experimental barns, where his grad student, Matthew Pelletier, is sitting on a wobbly chair, listening to country-western music and tapping on a little laptop plugged into the Automated Tomato Harvester, a beast as big as an 18-wheeler.
This is a famous machine. When its forerunner was introduced in the 1960s, it revolutionized tomato harvesting. Until then, all tomatoes were picked by hand mostly by Mexicans brought to California under the Bracero Program for migrant workers. They plucked the fruits, put them in boxes, hauled the heavy boxes to trucks, and away they went to the packing houses, which still produce more than 95 percent of all the salsa, ketchup, tomato soup, and pasta mix we eat in the United States today.
The tomato harvester changed forever the old way of doing business. The genetic engineers at Davis produced a "hard-skinned" tomato, and the mechanical engineers constructed the automated harvester, which grabs the whole plant with rubber "fingers" and shakes them up and down, causing all the green and red tomatoes to fall onto conveyer belts.
Over the years, many improvements have been made. Tomatoes are now sorted - green from red - aboard the harvester, weighed, and the useless vines shot out the back, sucked up by giant vacuum cleaners. Shrini, as he is known, envisions the tomato harvester of the future bristling with ever-smarter technologies: three or four sorters, yield monitors, graders, and yes, perhaps even a mold sensor.
Where once a dozen or more workers sorted tomatoes aboard the harvester, now there are only three or four. Soon there might not be any. Just a driver. And even he might be replaced.
Back in 1979, when the world seemed a much younger place, a group of small farmers and migrant worker activists, calling themselves the California Agrarian Action Project, sued the University of California for developing the harvester. They were called, dismissively, modern-day Luddites.
But they argued, forcefully and for a time successfully, that taxpayer dollars were being used to create labor-saving devices that put people out of work and benefitted only large operators with the money to buy the beasts. They won in the lower courts, but lost later on appeals. The progressive, former Texas state agriculture commissioner, Jim Hightower, wrote a book about the issue, called, Hard Tomatoes, Hard Times: The Failure of the Land Grant College Complex. It reads today, in the flood of digitized hype, like an ancient scroll.
But the case was a landmark, or a last gasp, depending on how one views the double-edged sword of technology. Due to it - and to the fact that there were so many migrant workers available at the time - researchers applying for federal grants for years had been dissuaded from going anywhere near the words "labor-saving device." Even today it is slightly odd to read the US Department of Agriculture's science publications and never see the word. Efficiency? Sure. Labor? Never.
But every researcher I spoke with at UC Davis, and at other schools around the country, admitted it. These technologies are designed, either explicitly or as a result, to take people out of the process. That might not be so bad. On the days I was in Davis, the temperatures soared above 104 degrees. Hand-picking tomatoes in July is hot, brutal, backbreaking work that pays a few dollars an hour. And farming remains, even today, one of the most dangerous jobs in America. Clearly, migrant farm workers are doing it because they desperately need the money. But I don't think it hurts to be honest about it. The farmers and the researchers want to replace men with machines.
Tony Turkovich is a real farmer. At his and his partner's family's 7,000-acre spread in Winters, California, he grows wheat, corn, alfalfa, beans, watermelons, cotton, prunes, oranges, and, of course, tomatoes. I drive the rental out across the warm fecund fields, eating a sloppy, wet, fresh roadside peach, with the windows down, radio cranked, hot and happy.
Tony is a professional. Conservative enough, but still young, and in the farm industry, he could fairly be called an "early adapter." He uses lasers to level his fields. Radar. Sensors. His desk is clean, almost free of paper. He has a laptop, which he uses for record-keeping, field data, accounting. He's getting his own weather station soon. He likes to test stuff in his fields for the engineers at UC Davis. He is our reality check.
We talk about the gizmos, their pros and cons. But this is the bottom line. If smart technologies work, if they can be priced affordably, farmers will buy every damn one of them they can. Tony says he and his neighbors simply cannot compete against a developing country when it comes to labor. Labor makes up 33 percent of his annual budget. It is too expensive in California. Too cheap in Nicaragua or Brazil. He is philosophical about it. He knows this means less men working his crops each year. But the ones that remain will inevitably, he believes, be better paid and better educated. Technicians, not field hands.
"I can see the number of employees being reduced year after year," he says. "To just a handful? I'm not so sure. Mother Nature throws a lot of variables at us, and things don't always go so smooth."
On the walls of his office, an old farm house - built by the descendants of the first pioneers to work this land nearly a hundred years ago - are old grainy black-and-white images. They show a train of mules pulling a combine. Then a steam engine. Then a gasoline-run tomato harvester.
We go outside, and he starts up his newest tractor; sitting in the air-conditioning cab, he flicks some switches, and onboard microprocessors come alive. This farmer is going digital. And everything we put in our bellies is going to be put there with the help of a computer.
