PASCO, Wash. — A retired Hanford Site mechanical engineer has for 11 years been seeking the holy grail of the apple industry — a robotic picker.
At least three companies envision robotics as an answer to the apple industry’s labor shortages and increasing labor costs.
“If the migrant worker thing blows, I could see a $3 billion (tree fruit) industry going up in smoke real quick. That would be devastating to the state and particularly Central Washington,” says Frank Moore, 71, who retired in 2006 after 33 years in design and application of remote systems for contractors at the Hanford Site, a nuclear production site that has been nearly completely decommissioned. The contractors included Westinghouse, Fluor and Duke Power.
What he’s done
Moore’s accomplishments included programming thermal hydraulics code for Babcock & Wilcox for the 1,000-megawatt nuclear power plant at Hanford known as WNP-1. He programmed the plant’s automated plutonium and uranium fuel lines and specialized in remote operations of fuel and radioactive material.
In the 1980s, Moore patented an automated nuclear fuel inspection system. He built the in-core materials test assembly for a test reactor to develop a breeder reactor program that was canceled by the Carter administration. He made cesium capsules for irradiating blood for surgery and managed a team building shipping casks and trailers for nuclear batteries for deep space probes.
He’s even patented a shock-absorbing hitch for tractor fork lifts to haul bins of fruit with less bouncing.
“I have a depth of experience in remote handling that no one else has that’s working on this (robotic apple picker). I have access to retired and working engineers I could utilize if I had the funding,” Moore says.
Working on the idea
Needing something to do upon retirement, he picked up the idea of an automated apple harvester that he began pondering in 1993 when his son-in-law owned apple and cherry orchards in Wenatchee.
“Labor was way too inexpensive back then and the electronic side was too expensive for it to work,” he said.
But those impediments has been reduced by 2006 when he retired and returned to the idea.
He looked at the intricacies and needs of apple harvest, including picking speed and gentle handling to avoid bruising. He decided suctioning apples off trees with vacuum tubes created too much velocity and increased bruising. His idea is an automated stem-cutting picker with a lower-velocity vacuum inside tubes to move apples to a decelerator and automated handling area where they are washed, culls are sorted out and good fruit goes into bins.
Some field sorting saves money at packing houses and washing removes sunburn protectant dust or film and chills the fruit, which can be important in hot weather, he said.
Decelerators have been bottlenecks in previous non-robotic harvest assist equipment, but Moore said he doesn’t think his decelerator, washing and sorting will slow picking because his design and technology are already proven in his system that inspects 6,000 nuclear fuel pellets per minute.
What really sets his system apart from others, he said, is early spring imaging and data collection. Imaging for leaf buds assists in pruning with a pruning head on the robotic arms. Later, with different heads, the robotic arms will clip stems to pick the fruit.
After pruning, a second round of imaging before leaves emerge logs the locations of fruit buds on a vector-based image.
“The big advantage of this is I have from April to August to generate all the pick programs for the robotics so it increases my speed significantly on going and finding fruit,” Moore said.
Pre-programming reduces the need for a lot of light to find fruit, so the picker can operate about 18 hours per day.
Moore received a patent priority date on Aug. 25, 2009, his patent in 2016 and created HarvestMoore LLC.
He is working to finish a crude prototype on his “hobby budget” but envisions a machine 15 to 18 feet long, 8 feet wide and with three to four robotic picking arms on each side to pick low, medium and high fruit on trees on both sides of a row.
The GPS-guided self-propelled machine will pick up empty bins in front and deposit full bins out the back and be capable of having one to three bins on board at a time. A low-profile model will work with V-trellis fruit tree systems.
His goal, like that of his competitors, is one apple picked per second per robotic arm because that’s about the speed of a human picker using both hands.
He believes his machine will be able to detect 99 percent of fruit buds but he has no field data on picking rate and bruising percentages because he hasn’t done much field testing.
He applied for a grant from the Washington Tree Fruit Research Commission but says Abundant Robotics of Hayward, Calif, beat him to it. The commission recently turned him down for a pruning grant.
Mike Willett, commission manager, declined to say why. Willett said the commission has given Abundant Robotics nearly $500,000 in grants and is strongly supportive of its work. He said he doesn’t know how realistic the 2018 or 2019 goals are for commercialization.
Moore also is trying to work with Oxbo International Corp., a Byron, N.Y., manufacturer of harvesters for seed corn, coffee beans, berries, peas, green beans, olives, citrus fruit and grapes.
“I want durable hydraulics. I like their hydraulic designs and see them as the best option. They are the No. 1 choice for me. I just have to convince them I’m the No. 1 choice for them,” Moore said.
With Oxbo’s help a full prototype could be built and tested next year with a goal of commercial release in 2019, he said.
Scott Korthuis, Oxbo berry and tree fruit product manager, said Oxbo is interested in Moore’s effort and thinks his fruit mapping system has merit but wants to see the machine work.
“We’ve given him some parts and hydraulics advice, so we are encouraging him but not jumping in feet-first,” Korthuis said.
Oxbo may buy Moore’s patent and company but it’s too early to know, he said.
“He hoped to have it running this (past) harvest, but it didn’t happen,” Korthuis said.
Oxbo spent at least $2 million building an apple harvest-assist machine several years ago but abandoned it because it wasn’t that much faster than good pickers on ladders, Korthuis said.
“I was only as fast as the slowest picker of four on the platform. One guy just doesn’t pick as fast or he runs into lots more apples than the other three. It’s hard to keep it productive,” he said.
While Abundant Robotics of Hayward, Calif., and FFRobotics in Israel say their goals are robotic pickers on the market by the 2018 fall harvest or early 2019, Korthuis said university experts say it’s “a ways off yet” because it is difficult for a robot to find an apple, reach out and pick it and get it into tubes or conveyors in one second without bruising.
Non-robotic shaker systems are used successfully for nuts and berries.
“We’ve solved the simple ones. It’s only the tough ones that are left like apples, strawberries and peaches,” Korthuis said.
Moore believes his system and those of Abundant Robotics and FFRobotics are viable, but his patent includes a large amount of detail covering picking and removing fruit. It covers mechanical aspects of the pruner and harvester including imaging, fruit deceleration and a unique system that locates each tree trunk by latitude and longitude.
He expects to get a second patent soon for his imaging system.
He believes the Israeli mechanical prongs picking fruit and placing it on a conveyor will be slower than clip picking and apple transport by vacuum tubes.
The three systems, he said, have yet to prove they are economically feasible at the harvest speeds that are needed.
“I think I have the best chance on economics,” Moore said. “It will cost $300,000 to $350,000 to build a six- to eight-arm machine and sell it for close to $500,000. That’s a lot so I have to have a pick rate that pays off in five years.”
Moore said he’s invested $100,000, mostly in attorney fees for patents and would like to recoup that and 50 percent of his hours if he were to sell his firm.
Origins: Born in Lakin, Kan., grew up on a cattle and wheat ranch in western Kansas.
Family: Wife, Nancy; two daughters; two grandchildren.