Driscoll’s is so secretive about its robotic strawberry picker it won’t let photographers within telephoto range of it.

But if you do get a peek, you won’t see anything humanoid or space-aged. AgroBot is still more John Deere than C-3PO — a boxy contraption moving in fits and starts, with its computer-driven sensors, graspers and cutters missing 1 in 3 berries.

Such has been the progress of ag-tech in California, where despite the adoption of drones, iPhone apps and satellite-driven sensors, the hand and knife still harvest the bulk of more than 200 crops.

Now, the $47-billion agriculture industry is trying to bring technological innovation up to warp speed before it runs out of low-wage immigrant workers.

California will have to remake its fields like it did its factories, with more machines and better-educated workers to labor beside them, or risk losing entire crops, economists say.

“California agriculture just isn’t going to look the same,” said Ed Taylor, a UC Davis rural economist. “You’re going to be hard-pressed to find crops grown as labor-intensively as they are now.”

Driscoll’s, which grows berries in nearly two dozen countries and is the world’s top berry grower, already is moving its berries to table-top troughs, where they are easier for both human and machines to pick, as it has done over the last decade in Australia and Europe.

“We don’t see — no matter what happens — that the labor problem will be solved,” said Soren Bjorn, president of Driscoll’s of the Americas.

That’s because immigrant farmworkers in California’s agricultural heartlands are getting older and not being replaced. After decades of crackdowns, the net flow across the U.S.-Mexico border reversed in 2005, a trend that accelerated through 2014, according to a Pew Research Center study. And native-born Americans aren’t interested in the job, even at wages that have soared at higher than average rates.

“We’ve been masking this problem all these years with a system that basically allowed you to accept fraudulent documents as legal, and that’s what has been keeping this workforce going,” said Steve Scaroni, whose Fresh Harvest company is among the biggest recruiters of farm labor. “And now we find out we don’t have much of a labor force up here, at least a legal one.”

Stated bluntly, there aren’t enough new immigrants for the state’s nearly half-million farm labor jobs — especially as Mexico creates competing manufacturing jobs in its own cities, Taylor said. He has calculated that the pool of potential immigrants from rural Mexico shrinks every year by about 150,000 people.

Not surprisingly, wages for crop production have climbed 13% from 2010 to 2015 — a higher rate than the state average, according to a Los Angeles Times analysis of Labor Department data.

Growers who can afford it have begun offering savings and health plans more commonly found in white collar jobs. And they’re increasingly turning to foreign guest workers, recruiting 11,000 last year, which is a fivefold jump in just five years, The Times found.

None of that will solve the problem, economists say. Changing what we grow and how we grow it is all that’s left.

Response has been uneven, at best. Vast areas of the Central Valley have switched from labor intensive crops such as grapes or vegetables to almonds, which are mechanically shaken from the tree. The high-value wine grape industry has re-engineered the bulk of its vineyards to allow machines to span the vines like a monorail and strip them of grape clusters or leaves.

Fresno’s raisin industry, however, has a tougher problem to solve on a tighter profit margin. To fully mechanize, it may have to change not just its vineyard design, but the grape variety itself, much like the tomato industry developed a tough skinned Roma to withstand mechanical harvesters.

When labor shortages and price shocks hit in the early 2000s, growers altered vineyards so that machines could shake partially withered Thompson seedless grapes onto paper trays, a method that can slash more than 80% of labor costs, according to U.C. Davis researchers.

Eliminating trays entirely, however, requires a grape that can dry slowly on the vine before September rains hit. Thompsons mature too late. The Sunpreme, developed by a retired USDA plant scientist in the Central Valley, may soon be widely available, said Matthew Fidelibus, a UC cooperative extension advisor.

It may be too late to mechanize asparagus. The crop, among the most labor-intensive in the state, has gradually shifted to Mexico since trade barriers made it cheaper to grow there, casting a nostalgic pall over Stockton’s asparagus festival.

Last year, farmers in the Sacramento-San Joaquin River Delta area harvested only 8,000 acres of the signature spear, which is depicted on water tanks and town emblems throughout the region. In 2000, they harvested 37,000 acres, according to the U.S. Department of Agriculture.

“We’re headed toward zero pretty soon,” said Cherie Watte Angulo, executive director of the California Asparagus Commission.

Grown on perennial beds that last a decade or so, asparagus must be selectively harvested every day during its 90-day season. Machines have utterly failed to duplicate human judgment and dexterity.

This season, a grower’s cooperative in the Stockton area tested a prototype harvester from Washington state that uses sensors to select only the mature stalks.

“I’ll keep it simple: the machine didn’t work,” said Bob Ferguson, who hosted the machine on his 162 acres of asparagus beds. “He took it back up to Washington.”

Even Driscoll’s AgroBot, among the more advanced prototypes in California fields, was picking only a bit more than half the ripe berries in its trials this spring in Camarillo.

“We think we are very close, but every day we try to make the next step. We see new things we need to solve,” said Juan Bravo, the Spanish inventor who is counting on Driscoll’s continued backing for his 10-year endeavor.

So far, the Watsonville, Calif., company is into AgroBot for the long-haul, said Michael Christensen, Driscoll’s research and development director, who watched Bravo tinker with the machine’s three dozen arms before setting it to another crawl.

Vertical rods slid left and right, guiding four-fingered graspers to precise coordinates set by a camera and computer. Soon, a stream of ripe berries emerged on a conveyor, mixed occasionally with green-tipped fruit.

“In some ways, you can look at it as every pound picked is part of the solution,” Christensen said.

The rest of the fruit industry has its eye on AgroBot’s trials, even as it looks to other start-ups such as Abundant Robotics, which hopes to duplicate the dexterity, judgment and perception of human apple pickers. Soft Robotics, based in Cambridge, Mass., boasts that its graspers can pick up a cupcake without damaging the icing.

Frank Maconachy is skeptical of solutions imported from tech centers. His company, Ramsay Highlander, started as a greasy machine shop in the Salinas Valley and slowly migrated toward Silicon Valley instead.

The company, with $15 million in annual sales, builds a fleet of computerized and sensor-driven machines for the lettuce and produce industry — and he is working with AgroBot’s U.S. competitor for strawberry picking, Harvest Croo, based in Plant City, Fla.

An early generation of robotic machine uses a band saw to mow whole rows of baby lettuce and other greens. But when produce giant Taylor Farms tried it on romaine heads, a slight height variation in the beds put the saw right across the heart of the heads, leaving nothing but shredded leaves, Maconachy said.

Maconachy developed a cutter using high-speed water jets. It now cuts all the romaine heads cleanly, and can be adapted for cabbage and celery.

“That machine took the work of 30 people and brought it down to about 12 people,” Maconachy said.

Cutting iceberg heads, especially large ones, remains problematic — it is planted so densely and the heads are so heavy it is difficult to maneuver cutters and graspers into beds. Maconachy thinks he has that engineering problem solved, but can’t raise the capital to develop it.

Ironically, plant scientists may have to reverse their cross-breeding to the original “iceberg” head, nicknamed from the tons of ice it took to keep it cool for cross-country train trips.

The crisphead variety used to be more bulb-shaped, which would give cutters and graspers more room to work, Maconachy said.

Rick Antle, chief executive of Tanimura & Antle, is whittling away at the labor on the planting side. He showed off his own robotic bet, called Planttape. The machine — equally homely as AgroBot — raced down a lettuce field outside Salinas, laying down a long strand of seedlings strung together on a bio-degradable tape, like 9-volt batteries in a 50-caliber machine gun belt.

 

That was twice the speed of its 35-year-old predecessor, and it required less than a tenth of the labor. To prove his point, Antle ran the old machine, which required three times the workers, on a nearby celery field. “That was it, for 35 years,” Antle said.

Lettuce growers usually plant seed, which can be unreliable, every few inches, then thin the field to fit the maximum number of heads at the optimal spacing. That means scores of workers in the spring have to walk row after row, moving inch by inch to pull seedlings over with a hoe — one of the oldest tools of agriculture.

The computer-guided See and Spray machine, developed by Silicon Valley start-up Blue River Technology, can do the work of 20 of those laborers before noon. It is one of five robotic thinners deployed on thousands of acres of summer lettuce in the Salinas Valley.

Diego Alctantar, 25, operated the tractor pulling See and Spray across a recently planted lettuce field near Gilroy. A computer guided jets of fertilizer-infused water to desiccate seedlings according to a kill-or-skip pattern that left nine-inch gaps between heads.

See and Spray

Alcantar, who grew up in the Salinas Valley, thinned lettuce and cut spinach the old way before getting his tractor license a few years ago. “It’s hard labor,” he said. “I have a lot of respect for them.”

The machine is not perfect. It can’t thin out two sprouts that come up together, and it misses some it is programmed to kill.

A small crew nearby strolled down the rows, taking care of stragglers with a few short hoe strokes.

Credits: Produced by Andrea Roberson